US20190130322A1 - Wait time prediction - Google Patents

Wait time prediction Download PDF

Info

Publication number
US20190130322A1
US20190130322A1 US15/800,787 US201715800787A US2019130322A1 US 20190130322 A1 US20190130322 A1 US 20190130322A1 US 201715800787 A US201715800787 A US 201715800787A US 2019130322 A1 US2019130322 A1 US 2019130322A1
Authority
US
United States
Prior art keywords
wait time
model
business
expected wait
information
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
US15/800,787
Inventor
Li-Sue Chen
Steve Chien
Quang Duong
Tong Wang
Xixi WANG
Ke Xu
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Google LLC
Original Assignee
Google LLC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Google LLC filed Critical Google LLC
Priority to US15/800,787 priority Critical patent/US20190130322A1/en
Assigned to GOOGLE LLC reassignment GOOGLE LLC ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: CHEN, LI-SUE, CHIEN, STEVE, DUONG, QUANG, WANG, TONG, WANG, Xixi, XU, KE
Publication of US20190130322A1 publication Critical patent/US20190130322A1/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063Operations research, analysis or management
    • G06Q10/0631Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10Complex mathematical operations
    • G06F17/18Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00Computing arrangements using knowledge-based models
    • G06N5/04Inference or reasoning models
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00Computing arrangements using knowledge-based models
    • G06N5/04Inference or reasoning models
    • G06N5/046Forward inferencing; Production systems
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/10Office automation; Time management
    • G06Q10/109Time management, e.g. calendars, reminders, meetings or time accounting
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00Commerce
    • G06Q30/02Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201Market modelling; Market analysis; Collecting market data
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00Commerce
    • G06Q30/02Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201Market modelling; Market analysis; Collecting market data
    • G06Q30/0202Market predictions or forecasting for commercial activities

Definitions

  • the DMV website shows wait time information at each of their offices. Certain hospitals and emergency rooms also publish their wait time information. Most establishments, however, do not collect or publish wait time information.
  • the present disclosure provides for receiving data related to a business, the data including at least historical wait time information, generating a model using one or more computing devices based on at least the historical wait time information for the business, determining an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model, receiving a request for information about a business, and providing for display, in response to the request, the expected wait time.
  • the present disclosure may further include requesting user feedback regarding the expected wait time, and refreshing the model as new wait times and other information are received.
  • Generating the model may include using other types of data in addition to historical wait times, such as place attributes, i.e., particular attributes about a business. Generating the model may also include using historical temporal signals, collected over a look-back window, that fluctuate throughout a day and throughout a week. The historical temporal signals may be further analyzed to derive useful statistics that can also be used in generating the model.
  • the model generated may be a regression model or a classification model, and may further be a linear model, a boosting tree model, a random forest model, or a neural net model.
  • Determining the expected wait time may include determining an upper bound expected wait time. Determining the expected wait time may further include determining whether the expected wait time is less than a predetermined significant-wait threshold. Determining the expected wait time may further include computing a maximum expected wait time for the particular day of the week, and determining at least one peak interval for the particular day of the week having the maximum expected wait time. Determining the expected wait time may also include smoothing the expected wait time data by taking weighted averages with expected wait times of the neighboring time periods.
  • Providing for display may further include displaying the time period as a no-wait time period if the expected wait time is less than the significant threshold.
  • Providing for display may also include displaying the maximum expected wait time of the particular day of the week and the corresponding peak interval.
  • Providing for display may yet include displaying in association with other business information indicating busyness level.
  • the disclosure further provides a system, comprising a memory and one or more processors in communication with the memory.
  • the one or more processors are configured to receive data related to a business, the data including at least historical wait time information, generate a model using one or more computing devices based on at least the historical wait time information for the business, determine an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model, receive a request for information about a business, and provide for display, in response to the request, the expected wait time.
  • the disclosure yet further provides a computer-readable storage medium storing instructions executable by one or more processors for performing a method, comprising receiving data related to a business, the data including at least historical wait time information, generating a model using one or more computing devices based on at least the historical wait time information for the business, determining an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model, receiving a request for information about a business, and providing for display, in response to the request, the expected wait time.
  • FIG. 1 illustrates an example of how expected wait time information may be displayed according to aspects of the disclosure.
  • FIG. 2 is a block diagram illustrating an example system for predicting wait times according to aspects of the disclosure.
  • FIG. 3 is a flow diagram illustrating an example method of generating a wait time prediction model, determining expected wait times using the model, and providing the expected wait times for display in accordance with aspects of the disclosure.
  • FIG. 4 is a flow diagram illustrating an example method of generating a wait time prediction model by training computing devices to learn a pattern from various training data in accordance with aspects of the disclosure.
  • FIG. 5 is another flow diagram illustrating an example of optimizing the predicted wait time for display according to aspects of the disclosure.
  • FIG. 6 is another flow diagram illustrating an example method for determining a maximum expected wait time providing the maximum expected wait time for display according to aspects of the disclosure.
  • FIG. 7 illustrates example displays of expected wait time information according to aspects of the disclosure.
  • the technology relates generally to predicting and displaying wait times likely to be experienced by users visiting a business or a public service establishment.
  • the technology assists users in making everyday decisions, for example, which grocery store to stop by in the next 30 min; which restaurants nearby have a short wait time for dinner around 7 pm tomorrow; when would be the best time to try that very popular restaurant in town to avoid a very long wait?
  • the expected wait time information may be provided for display.
  • map information related to the business is provided, along with other information corresponding to the business. Examples of such other information include a phone number, directions, website, operating hours, etc.
  • Example of other related information may include a graph showing how popular the business is at various times of a day on various days of a week, relative to the typical peak popularity for the business for the week (“popular times”).
  • the current expected wait time may be provided for display.
  • the expected wait time for each time period may be overlaid with the popular times graph.
  • the maximum expected wait time for a particular day may be provided for display, along with peak intervals that correspond to the maximum wait time.
  • the wait time prediction technology described herein builds a model based on historical wait times collected from users. As more new wait times are collected, the model may be refined or refreshed. This allows the model to be fine-tuned over time. For example, as more data points are collected, the impact of outliers will be reduced. As another example, new data points can also show certain changes in the wait time patterns for a business (e.g., maybe a place is gaining popularity).
  • the wait time prediction technology may additionally use other types of data in building and refining the model, such as place attributes, i.e., particular attributes about a business. For example, wait time patterns are likely very different for a restaurant as compared to a hospital on a weekend, for a grocery store as compared to a department store on a weekday, or for a fine dining restaurant as compared to a fast food restaurant during lunch hours. Therefore, place category may be play an important role in the wait time pattern. Another important place attribute is whether the business accepts or recommends making a reservation. Yet another place attribute that may play a role in the wait time pattern is the popularity of the business, for example, as reflected by user ratings and comments. Other place attributes, such as online search and visit statistics, may also play a role in the wait time patterns and therefore used in building the prediction model.
  • place attributes i.e., particular attributes about a business. For example, wait time patterns are likely very different for a restaurant as compared to a hospital on a weekend, for a grocery store as
  • the wait time prediction technology may further use historical temporal signals that fluctuate throughout a day and throughout a week to build and refine the prediction model. For example, visit data such as the number of arrivals, the number of departures, the occupancy level, and the duration of visits at a particular time period and a particular day of the week can be collected for a predetermined look-back window and used to predict patterns. Other examples include online visits and searches that fluctuate depending on the time of the day and the day of the week.
  • the historical temporal signals described above may be further analyzed over a part of or the entire look-back window to derive useful statistics that can further help predicting wait times at a business, such derived temporal signals may also be used to build and refine the wait time prediction model. For example, a maximum, a minimum, a mean, and a sum of the number of arrivals at a particular time period on a particular day of a week for a business may be computed over half of the look-back window. As another example, aggregated data such as the 90 th percentile visit duration at a particular time period on a particular day of a week for a business can be computed over the entire look-back window.
  • lag variables of each of the historical temporal signals and derived temporal signals described above can be determined.
  • the lag variable may be backward-looking or forward-looking.
  • one interesting backward-looking lag variable may be the number of arrivals in the time period preceding the time period for which an expected wait time is to be predicted, since these recent arrivals are likely to remain for more than one time period.
  • one interesting forward-looking lag variable may be the occupancy level in the upcoming time period following the time period for which an expected wait time is to be predicted. For instance, if the following hour corresponds to the end of lunch break for many employees in the area, a large number of departures may be expected in the upcoming hour, and therefore the business will soon be able to absorb more customers in the upcoming hour.
  • the model may be, for example, a regression model or a classification model.
  • the model may be a linear model, a boosting tree model, a random forest model, or a neural net model.
  • a quantile regression model is used to determine an upper bound wait time for the given time period. For example, the 90 th percentile expected wait time.
  • a deep neural network with a custom loss function may be used to implement the quantile regression model.
  • the expected wait time may be determined and expressed as an upper bound wait time. For example, “up to 30 min wait” is displayed to the user.
  • the expected wait time may be rounded to a predetermined granularity for display, e.g., instead of “14 min wait” and “36 min wait,” the users will see “15 min wait” and “30 min wait.”
  • the wait time is smoothed by taking a weighted average of the expected wait time and at least one other expected wait time from a neighboring time period.
  • a certain threshold of wait time is almost always expected for a category of businesses. For example, a person walking into almost any sit-down restaurant could expect at least a trivial wait to be seated, say, 10 min. Thus, it would not be interesting for a user to see such trivial wait times. Therefore, in one example, if for a given time period, the expected wait time is less than a predetermined significant-wait threshold, then “usually no wait” may be displayed for the given time period.
  • the wait time prediction technology determines and displays a maximum expected wait time for the day.
  • the peak interval including all the time periods that correspond to the maximum expected wait time may be displayed. For example, “longest wait time today ⁇ 60 min ⁇ 12:30 pm-2 pm, 5:00 pm-7:30 pm.” If the expected wait time is less than the predetermined significant-wait threshold for all time periods of a day, then “usually no wait” may be displayed instead of the maximum expected wait time.
  • feedback is requested from the same user who requested the wait time information.
  • the user can, for example, provide feedback about the accuracy of the expected wait time, input a new wait time actually experienced by the user, or provide additional information about the business or about the user's visit. Such information may be used to refine and refresh the wait time prediction technology.
  • the wait time prediction technology described herein is advantageous in a number of ways. For example, because a larger amount of historical data can be collected as compared to requesting users currently visiting a business for real-time data, the predicted wait times will be less affected by outliers and more indicative of a trend. Further, because the wait time prediction model can be refined and refreshed with newly collected data, the wait time prediction technology can be updated to reflect a recent change in trend.
  • the data used to build the wait time prediction model may be large in volume, but the users are presented with useful information that is simple to understand. Therefore, after a quick glance, users will be aided in their decision making.
  • FIG. 1 illustrates an example of how the expected wait time information may be displayed.
  • graphic 100 includes a map section 110 , an informational section 120 , and a popular times section 130 .
  • the graphic 100 may be displayed, for example on a client device, in response to a request for information related to Business A.
  • the request for information may include, for example, an address, business name, general geographical area, type of business, etc.
  • the map section 110 may include a depiction of geographical objects at a particular geographic location.
  • the geographic objects may include roads, buildings, landmarks, statues, street signs, etc.
  • the objects may be depicted in, for example, a roadgraph, aerial imagery, street level imagery, or the like.
  • the map section 110 may also depict a marker 112 corresponding to a requested location, such as Business A.
  • the marker identifies where the requested location is located with respect to other objects depicted in the map section.
  • the marker may be depicted in any of a variety of shapes and forms.
  • a label 114 may identify that the marker 112 corresponds to Business A. Though not shown, in some examples the marker or map near the marker may be depicted so as to indicate a current expected wait time at Business A.
  • a size, shape, shading, or other aspect of the marker or map may change if the current expected time is short or long.
  • additional icons may be positioned on the map near the marker to indicate current expected wait time. It should be understood that these are merely examples, and any of a number of indicators may be used.
  • Informational section 120 provides further information related to Business A. Such information may include user input controls 122 , 124 , 126 , enabling the user to take a predetermined action.
  • the predetermined action may be, for example, calling a phone number for Business A, requesting navigational information to Business A, or visiting another website such as a site for Business A.
  • the informational section 120 may also include text 128 , such as a listing of the operating hours of Business A.
  • the popular times section 130 includes a graphic representation of how popular the requested location is throughout the day and the week.
  • a popular times histogram 131 for the current day is shown, the user may select another day 132 to view histograms for other days.
  • typical levels of busyness are indicated by unshaded bars 133
  • the current level of busyness, in real-time is indicated by a shaded bar 134 .
  • the day is broken down on an hourly basis.
  • the expected wait time 135 for the current hour is displayed as a text above the popular times histogram 131 .
  • Below the popular times histogram 131 the maximum expected wait time 136 for the current day, along with the corresponding peak interval, is shown as a text.
  • the user can click on the popular times histogram 131 to see the expected wait time for a different hour, and by selecting a different day to view on the popular times section 130 , can view the expected wait times for a different day.
  • the timeline 137 may correspond to a range of hours that Business A is in operation.
  • the expected wait time information may be provided for display in any of a number of ways, such as text, pictorial diagrams, charts, graphs, etc. In other examples, the expected wait time information may be provided to other applications, such as scheduling applications, ride-service applications, games, or any of a variety of other applications.
  • FIG. 2 illustrates an example system used to receive data related to a business, generate a wait time prediction model, determine expected wait times using the generated model, and provide the expected wait times for display upon user request.
  • system 200 can include computing devices 210 in communication with one or more computing devices 260 , 270 , as well as storage system 240 , through network 250 .
  • Each of the computing devices 210 contains one or more processors 220 , memory 230 and other components typically present in general purpose computing devices.
  • Memory 230 of each of the computing devices 210 can store information accessible by the one or more processors 220 , including instructions 234 that can be executed by the one or more processors 220 .
  • Memory 230 can also include data 232 that can be retrieved, manipulated or stored by the processor.
  • the memory can be of any non-transitory type capable of storing information accessible by the processor, such as a hard-drive, memory card, ROM, RAM, DVD, CD-ROM, write-capable, and read-only memories.
  • the instructions 234 can be any set of instructions to be executed directly, such as machine code, or indirectly, such as scripts, by the one or more processors.
  • the terms “instructions,” “application,” “steps,” and “programs” can be used interchangeably herein.
  • the instructions can be stored in object code format for direct processing by a processor, or in any other computing device language including scripts or collections of independent source code modules that are interpreted on demand or compiled in advance. Functions, methods, and routines of the instructions are explained in more detail below.
  • Data 232 may be retrieved, stored, or modified by the one or more processors 220 in accordance with the instructions 234 .
  • the data can be stored in computer registers, in a relational database as a table having many different fields and records, or XML documents.
  • the data can also be formatted in any computing device-readable format such as, but not limited to, binary values, ASCII or Unicode.
  • the data can comprise any information sufficient to identify the relevant information, such as numbers, descriptive text, propriety codes, pointers, references to data stored in other memories such as at other network locations, or information that is used by a function to calculate the relevant data.
  • the one or more processors 220 can be any conventional processors, such as a commercially available CPU. Alternatively, the processors can be dedicated components such as an application specific integrated circuit (“ASIC”) or other hardware-based processor. Although not necessary, one or more of the computing devices 210 may include specialized hardware components to perform specific computing processes, such as decoding video, matching video frames with images, distorting videos, encoding distorted videos, etc. faster or more efficiently.
  • ASIC application specific integrated circuit
  • FIG. 2 functionally illustrates the processor, memory, and other elements of computing devices 210 as being within the same block
  • the processor, computer, computing device, or memory can actually comprise multiple processors, computers, computing devices, or memories that may or may not be stored within the same physical housing.
  • the memory can be a hard drive or other storage media located in housings different from that of the computing devices 210 .
  • references to a processor, computer, computing device, or memory will be understood to include references to a collection of processors, computers, computing devices, or memories that may or may not operate in parallel.
  • the computing devices 210 may include server computing devices operating as a load-balanced server farm, distributed system, etc.
  • some functions described below are indicated as taking place on a single computing device having a single processor, various aspects of the subject matter described herein can be implemented by a plurality of computing devices, for example, communicating information over network 250 .
  • Each of the computing devices 210 , 260 , 270 can be at different nodes of a network 250 and capable of directly and indirectly communicating with other nodes of network 250 .
  • FIG. 2 Although only a few computing devices are depicted in FIG. 2 , it should be appreciated that a typical system can include a large number of connected computing devices, with each different computing device being at a different node of the network 250 .
  • the network 250 and intervening nodes described herein can be interconnected using various protocols and systems, such that the network can be part of the Internet, World Wide Web, specific intranets, wide area networks, or local networks.
  • the network can utilize standard communication protocols, such as Ethernet, WiFi and HTTP, protocols that are proprietary to one or more companies, and various combinations of the foregoing.
  • each of the computing devices 210 may include web servers capable of communicating with storage system 240 as well as computing devices 260 , 270 via the network 250 .
  • server computing devices 210 may use network 250 to transmit and present information to a user on a display, such as display 265 of computing device 260 .
  • computing devices 260 , 270 may be considered client computing devices and may perform all or some of the features described herein.
  • Each of the client computing devices 260 , 270 may be configured similarly to the server computing devices 210 , with one or more processors, memory and instructions as described above.
  • Each of the client computing devices 260 , 270 may be a personal computing device intended for use by a user, and have all of the components normally used in connection with a personal computing device such as a central processing unit (CPU) 262 , memory (e.g., RAM and internal hard drives) storing data 263 and instructions 264 , a display such as display 265 (e.g., a monitor having a screen, a touch-screen, a projector, a television, or other device that is operable to display information), and user input device 266 (e.g., a mouse, keyboard, touch-screen, or microphone).
  • CPU central processing unit
  • memory e.g., RAM and internal hard drives
  • display 265 e.g., a monitor having a screen, a touch-screen, a projector, a television, or other device that
  • the client computing device may also include a camera 267 for recording video streams and/or capturing images, speakers, a network interface device, and all of the components used for connecting these elements to one another.
  • the client computing device 260 may also include a location determination system, such as a GPS 268 .
  • Other examples of location determination systems may determine location based on wireless access signal strength, images of geographic objects such as landmarks, semantic indicators such as light or noise level, etc.
  • client computing devices 260 , 270 may each comprise a full-sized personal computing device, they may alternatively comprise mobile computing devices capable of wirelessly exchanging data with a server over a network such as the Internet.
  • client computing device 260 may be a mobile phone or a device such as a wireless-enabled PDA, a tablet PC, a netbook, a smart watch, a head-mounted computing system, or any other device that is capable of obtaining information via the Internet.
  • the user may input information using a small keyboard, a keypad, microphone, using visual signals with a camera, or a touch screen.
  • storage system 240 can be of any type of computerized storage capable of storing information accessible by the server computing devices 210 , such as a hard-drive, memory card, ROM, RAM, DVD, CD-ROM, write-capable, and read-only memories.
  • storage system 240 may include a distributed storage system where data is stored on a plurality of different storage devices which may be physically located at the same or different geographic locations.
  • Storage system 240 may be connected to the computing devices via the network 250 as shown in FIG. 1 and/or may be directly connected to any of the computing devices 210 .
  • Storage system 240 may store data, such as historical wait time information for particular businesses, place attributes, historical temporal signals, thresholds, etc. For example, the historical wait time information for a popular business for a past number of months or years may be stored.
  • the computing devices 210 may generate a wait time prediction model for a business and use the model to determine expected wait times at various time periods of a day throughout a week for the business.
  • the computing devices 210 may also receive requests from users' client computing devices 260 , 270 about the business and in response, provide for display on the client computing devices 260 , 270 the expected wait time at the business.
  • example methods are now described. Such methods may be performed using the systems described above, modifications thereof, or any of a variety of systems having different configurations. It should be understood that the operations involved in the following methods need not be performed in the precise order described. Rather, various operations may be handled in a different order or simultaneously, and operations may be added or omitted.
  • FIG. 3 illustrates an example method 300 of receiving data related to a business, generating a wait time prediction model, determining expected wait times using the model, and providing the expected wait times for display in response to a request.
  • data related to a business is received.
  • the data may be collected in a number of ways, for example, from databases, from the businesses themselves, from user reports, from real-time measurements, etc.
  • Example data may include the historical wait times experienced at the business at various times of the day, and on different days of the week.
  • the data may also include place attributes, i.e., information about the business, such as the category of the business, reservation information, whether there is usually a wait, online visit statistics, and user ratings and reviews, etc.
  • the data may further include historical temporal signals such as the number of arrivals, number of departures, occupancy level, and duration of visit at different times.
  • the data may also include signals that are derived from such historical temporal signals, for example, a maximum, a minimum, a mean, a sum, a percentile, etc.
  • the data may yet further include lag variables of the historical temporal signals and derived temporal signals. Examples of the data received are described in more detail in FIG. 4 .
  • a model is generated by training computing devices to learn patterns from the received data.
  • a deep neural network with a custom loss function may be trained with the received data to generate a quantile regression model by minimizing the loss function.
  • the neural network may have two fully connected hidden layers, each with tens or hundreds of neurons.
  • the model may be a regression model or a classification model.
  • the model may also be a linear model, a boosting tree model, a random forest model, or a neural net model.
  • One example of generating a model is described in more detail in FIG. 4 .
  • the expected wait time is determined using the model generated in block 310 . For example, when various place attributes and historical temporal signals about a business without known historical wait times are given to the generated model, expected wait times may be computed for the business using the model.
  • the determining may also include finding an upper bound expected wait time.
  • the determining may further include determining whether the expected wait time is below a predetermined significant-wait threshold.
  • the determining may also include computing the maximum expected wait time for the day.
  • the determining may yet further include smoothing the expected wait time using weighted averages.
  • One example of determining the expected wait time is described in more detail in FIG. 5 and FIG. 6 .
  • a request is received from a user for information on the business.
  • an example request could be an online search for a particular business, a map search for a particular business, a general search for a category of businesses nearby, or the like.
  • the expected wait time is provided for display to the requesting user.
  • the providing for display step may include, for example, providing for display the wait time as an upper bound, providing for display a time period as “no wait” if the expected wait time is below a significant-wait threshold, providing for display the maximum wait time of the day and the corresponding peak intervals, providing for display feedback options, and providing for display expected wait time information in association with other busyness information.
  • the expected wait time information may be provided for display in any of a number of ways, such as text, pictorial diagrams, charts, graphs, etc. In other examples, the expected wait time information may be provided to other applications, such as scheduling applications, ride-service applications, games, or any of a variety of other applications. Some example displays are described in more detail in FIG. 7 .
  • a request for feedback may include, for example, confirmation of whether the displayed expected wait times are consistent with the user's experienced wait time.
  • request for feedback is described in more detail in FIG. 7 .
  • the model is updated using new information received from user feedback.
  • new information may be newly experienced wait times, new reservation information, new arrival count at a certain hour, etc.
  • the new information may be used to refresh the training data for generating the model.
  • One example of updating the model is described in more detail in FIG. 4 .
  • FIG. 4 illustrates in more detail an example method 400 for generating a wait time prediction model.
  • some of the steps in the method 400 may correspond to blocks 310 , 320 , and 370 .
  • Block 410 example features for generating the model are collected and stored.
  • the data may be collected in a number of ways, for example, from databases, from the businesses themselves, from user reports, etc.
  • Block 420 shows that metadata may be an example feature data that is collected and stored.
  • the metadata could include the time period of the day and the day of the week. In this example, a week is divided into 648 equal time periods, each 30 minutes in length, e.g., Monday 8 am-8:30 am, Monday 8:30 am-9 am, . . . , Sunday 9:30 pm-10 pm, etc.
  • the metadata may also include the amount of time since the first opening hour of the business, and the amount of time until the last opening hour of the business.
  • place attributes about the business may also be an example feature data that is collected and stored.
  • the data may be collected in a number of ways, for example, from databases, from the businesses themselves, from user reports, etc.
  • Some example place attributes may include the category of the business, number of online visits, reservation information, user review information, etc.
  • the categories could be general, e.g., restaurant, store, hospital, government, etc.; or they could be more nuanced, e.g., fine-dining, fast food, bar, coffee shop, grocery store, department store, clinic, urgent care, post office, DMV, etc.
  • the place attributes may also include online visit statistics, such as number of times an ad about the business is shown as a result of a search, number of online searches about the business, number of online visits of the business's website, number of map searches, etc.
  • the place attributes may additionally include reservation information, such as whether the place accepts reservations, whether it recommends reservations at certain times.
  • User review and rating information may also be place attributes, such as the number of reviewers, the number of comments, the average rating, the number of ratings, etc.
  • Yet another example place attribute is whether the business is known to usually have a wait.
  • Block 440 shows that historical temporal signals may be another example feature data that is collected and stored.
  • Historical temporal signals are signals that may fluctuate throughout a day and throughout a week, therefore, they are collected for each (p, t) tuple, where p is the identifier of the place or the business, and t is the time period for which an expected wait time is to be determined.
  • the look-back window represents the number of data points collected for each (p, t) tuple. For example, if the look-back window is 24 weeks, then there would be 24 data points for each (p, t) tuple. In other words, this means that for Business A at the 8 pm-8:30 pm time period, there are 24 arrival counts recorded over these 24 weeks.
  • the look-back window can be tailored such that a representative amount of historical temporal signals are available, yet not look so far back such that there may have been significant changes in the wait time pattern for the business.
  • Example historical temporal signals include an arrival count, a departure count, an occupancy level, and durations of visits that started in the time period t.
  • the duration of visit data may be the mean duration, the 10 th percentile duration, the 90 th percentile duration, or the like.
  • Historical temporal signals may also include online behavioral signals, such as the number of times an ad about the business is shown as a result of a search, the number of online searches about the business, the number of online visits of the business's website, the number of map searches, etc.
  • Block 442 shows that derived temporal signals may be yet another example feature data that is collected and stored. For example, a maximum, a minimum, a mean, and a sum of the number of arrivals at a particular time period on a particular day of a week for a business, i.e., a (p, t) tuple, can be computed over the entire look-back window. The same set of statistics can be computed for each of the historical temporal signals described above in relation to block 440 . As another example, aggregated data such as the 10 th percentile and 90 th percentile visit durations for a (p, t) tuple can be computed over the entire look-back window. Instead of using the entire look-back window, alternatively one may also compute derived temporal signals over only part of the look-back window.
  • a maximum, a minimum, a mean, and a sum of the number of arrivals at a particular time period on a particular day of a week for a business i.e.
  • Block 444 shows that lag variables, yet another example of derived temporal signals, may also be a feature data that is collected and stored.
  • Lag variables may be determined for each of the historical temporal and derived temporal signals shown in blocks 440 and 442 .
  • a lag variable may be backward-looking or forward-looking.
  • one interesting backward-looking lag variable may be the number of arrivals in an earlier time period of the day, such as in the time period t ⁇ 1 preceding the time period t for which an expected wait time is to be predicted, since these recent arrivals are likely to remain beyond the time period t.
  • one interesting forward-looking lag variable may be the occupancy level in a later time period of the day, such as in the upcoming time period t+1 following the time period t for which an expected wait time is to be predicted. For instance, if the following hour corresponds to the end of lunch break for many employees in the area, a large number of departures may be expected in the upcoming hour, and therefore the business will soon be able to absorb more customers in the upcoming hour.
  • example labels for generating the model are collected and stored.
  • Block 460 shows that historical wait time data may be an example label data that is collected and stored.
  • the historical wait time data may be collected from consumer surveys. For example, the survey may ask a user to input the actual wait time experienced, or it may ask a user to choose from a list of estimates, e.g., 0 min, 5 min, 10 min, . . . , 30 min, 60 min, more than 60 min, etc. The survey may also ask the user at what time and day the wait time was experienced. The survey may additionally ask the user whether service was eventually received at the business. Additionally, the survey may ask more nuanced questions such as whether a reservation was made, was it dine in or take out, etc. To ensure that the received wait time data are accurate, the surveys are preferably sent to the user soon after the visit. The method is not limited to using surveys, for example, the actual wait time data may be collected from another database, from the businesses, or from real-time measurements, etc.
  • the historical wait time data collected are sorted and stored for each (p, t) tuple. For example, if a wait time is collected for a visit to Business A at 5:13 pm on Monday, it will be sorted as a wait time for Business A, Monday 5 pm-5:30 pm.
  • a deep neural network with a custom loss function is trained with the example features collected and stored in block 410 and the example labels collected and stored in block 450 .
  • the neural network may have two fully connected hidden layers, each with tens or hundreds of neurons.
  • the deep neural network learns patterns from the feature data and label data and generates a quantile regression model by minimizing the loss function.
  • the quantile regression model predicts an expected wait time as a value at a predetermined quantile, specifically, the 90 th percentile expected wait time.
  • the method 400 is not limited to training a neural network to generate a quantile regression model.
  • Various types of models may be generated using the method 400 .
  • the model may be a regression model or a classification model.
  • the model may also be a linear model, a boosting tree model, a random forest model, or a neural net model, etc.
  • the training data are refreshed with feedback collected from users.
  • the user who requested the wait time information may provide feedback about the accuracy of the expected wait time, or input a new wait time actually experienced by the user.
  • the feedback may include additional options such as leave a review or a rating, or provide more information about the business.
  • Feedback may additionally be collected via other methods and from other sources other than the requesting user, for example, they could be collected from new customer surveys, databases, etc.
  • the feedback may include a newly experienced actual wait time, which becomes an additional set of label data that can be used to refine and update the current model.
  • the feedback may also include new information that a user provides about the business, for example, that the place recommends reservations for lunch. Therefore, the current model may be updated with the new place attribute about reservations.
  • the feedback may also include a new duration of visit that the user reports, which can be used to update the historical temporal signals.
  • the model may be updated each time a new feedback is collected, or alternatively and more practically, for example, at some regular time interval, whenever a predetermined number of new feedbacks are collected, or whenever a new place attribute needs to be updated, etc.
  • FIG. 5 illustrates in more detail an example method 500 for determining an expected wait time using the model generated from the method 400 and providing for display.
  • some of the steps in the method 500 may correspond to blocks 330 and 350 .
  • Block 510 shows that an upper bound expected wait time is determined using the model generated from method 400 .
  • An upper bound expected wait time may be more useful for decision making than, for example, a lower bound or the mean expected wait time, since a user may have limited availability, e.g., lunch hour ends.
  • the absolute upper bound one preferred example determines the 90 th percentile expected wait time.
  • the upper bound could be the 70 th percentile expected wait time, and so on.
  • the method is not limited to expressing the expected wait time as an upper bound, for example, the expected wait time may also be expressed as a range of typically expected wait times or the most likely expected wait time.
  • the expected wait time is rounded to a predetermined granularity. For example, if granularity is set at 15 min, “up to 30 min wait” and “up to 45 min wait” may be provided for display instead of “up to 28 min wait” and “up to 40 min wait.” Continuing from the example above for the time period Friday 5:30 pm-6 pm, the expected wait time of 50 min may be rounded to 45 min.
  • the expected wait time is compared to a significant-wait threshold.
  • a certain threshold of wait time is almost always expected for a category of businesses. For example, a person walking into almost any sit-down restaurant without a reservation could expect at least some wait before seated, say, 10 min. Thus, it would not be interesting for a user to see such trivial wait times. Since upper bound expected wait times are determined in this example, a significant-wait threshold of 30 min may be preferable. Therefore, continuing from the examples above, for Friday 5 pm-5:30 pm the expected wait time is less than 30 min, so “usually no wait” will be provided for display (block 570 ) for Friday 5 pm-5:30 pm. For Friday 5:30 pm-6 pm in the example above, the expected wait time is greater than 30 min, therefore “up to 45 min wait” will be provided for display (block 580 ) for Friday 5:30 pm-6 pm.
  • FIG. 6 illustrates in more detail an example method 600 for determining a maximum expected wait time using the model generated from the example method 400 and providing the maximum expected wait time for display.
  • some of the steps in the method 600 may correspond to blocks 330 and 350 .
  • a maximum expected wait time for the day is determined among all the expected wait times available for the day.
  • the maximum expected wait time of the day is less than the predetermined significant-wait threshold. If so, “usually no wait” is provided for display for the day 630 .
  • the threshold for determining whether there is a significant wait for a particular time period is the same as the threshold for determining whether there is a significant wait for a day; however, alternatively, two different thresholds can be set.
  • the peak interval corresponding to the maximum expected wait time is determined. In this example, it is the longest interval including all the time periods that have the maximum expected wait time for the day. For example, if Friday from 7 pm-8:30 pm, the expected wait time is at maximum for the day at 90 min, then 7 pm-8:30 pm would be the peak interval for Friday, which includes all three time periods 7 pm-7:30 pm, 7:30 pm-8 pm, and 8 pm-8:30 pm.
  • the method may determine multiple peak intervals for a given day.
  • the peak interval is only 30 min in length. If the peak interval is greater than 30 min in length, as in the example above, the maximum expected wait time along with the corresponding peak interval will be provided for display (block 660 ). If the peak interval is only 30 min in length, it is determined whether any of its neighbors have an expected wait time that is only one granularity less than the maximum expected wait time (block 670 ). If so, such neighbors will be included in the peak interval (block 680 ) and the expected wait time of the entire peak interval will be reset to the majority value of the peak interval (block 690 ).
  • the peak interval is only 30 min in length.
  • the expected wait time is 30 min for both 6:30 pm-7 pm and at 7:30 pm-8 pm on Thursday. Since the granularity is set at 15 min, the expected wait times for both 6:30 pm-7 pm and 7:30 pm-8 pm are only one granularity below the maximum expected wait time of 45 min. Therefore, the new peak interval will be 6:30 pm-8 pm on Thursday, and the expected wait time for this new peak interval will be the majority value of 30 min.
  • FIG. 7 illustrates example displays 700 of the expected wait time data.
  • the current expected wait time data is displayed along with the popular times data, including real-time busyness data.
  • a user sent a request for information about the business at 7:12 pm on a Wednesday, therefore the expected wait time 710 for the time period 7 pm-7:30 pm on a Wednesday is shown.
  • the expected wait time 710 is expressed as an upper bound, “up to 45 min wait.”
  • the text “Correct?” links the requesting user to a feedback form, where the user may provide feedback about whether the information is accurate, an actually experienced wait time, additional information about the business, or other information related to the user's visit.
  • Pop-up 720 shows that, when the user clicks on the bar representing the time period 8 pm-9 pm on Wednesday, the higher expected wait time between the two time periods 8 pm-8:30 pm and 8:30 pm-9 pm is displayed.
  • the expected wait time for 8 pm-8:30 pm is 15 min and the expected wait time for 8:30 pm-9 pm is 0 min. Therefore, the expected wait time for the time period 8 pm-9 pm on Wednesday is expressed as “usually no wait” since both values are below the predetermined significant-wait threshold.
  • the threshold is set at 30 minutes.
  • a maximum expected wait time 730 for the current day is displayed along with the corresponding peak time interval.
  • the maximum expected wait time 730 for Wednesday is displayed as “longest wait today ⁇ 45 min ⁇ 5:30 pm-7 pm.”
  • the text “Useful?” links the user to a feedback form, where the user may provide feedback about whether the information is accurate, provide an actually experienced wait time, additional information about the business, or other information related to the user's visit.
  • Pop-up 740 shows that, when the user changes the date on the dropdown menu for dates 750 to Monday, the expected wait time information for Monday is displayed.
  • Monday's maximum expected wait time 760 is expressed as “usually no wait” because there is no time period on Monday with an expected wait time equal or above the predetermined significant-wait threshold, which is set at 30 minutes in this example.
  • the method may provide for display “recommend reservations” instead of “usually no wait,” as the latter may be misleading.
  • the examples of how the predicted wait times are displayed are not limiting, and that the information may be displayed in any of a variety of other formats.
  • the expected wait time information may be provided for display as text, pictorial diagrams, charts, graphs, etc.
  • the expected wait time information may be provided to other applications, such as scheduling applications, ride-service applications, games, or any of a variety of other applications.
  • the wait time prediction technology described herein is advantageous in a number of ways.
  • the technology may be developed for a large scale of businesses and public service establishments, and therefore can be relied on by users in making many everyday decisions.
  • the predicted wait times will be less affected by outliers and more indicative of a trend.
  • the wait time prediction model can be refined and refreshed with newly collected data, the wait time prediction technology can be updated to reflect a recent change in trend.
  • the data used to build the wait time prediction model may be large in volume, but the users are presented with useful information that is simple to understand. Therefore, after a quick glance, users will be aided in their decision making.

Abstract

The wait time prediction technology determines expected wait times for businesses or other public services using a model generated based on at least historical wait times for the business. In response to a request from a user, an expected wait time for service at the business for at least one particular time period on a particular day of a week is determined using the model and provided for display. User feedback regarding the expected wait time may be requested, and used to refresh the model as new wait times and other information are collected.

Description

    BACKGROUND
  • In making everyday decisions, it is often helpful to know how busy a business or public service establishment is before planning a visit. For example, the DMV website shows wait time information at each of their offices. Certain hospitals and emergency rooms also publish their wait time information. Most establishments, however, do not collect or publish wait time information.
  • Currently, some websites can indicate how busy any of a variety of businesses or public establishments typically is at various times of the week or in real-time. However, this only provides measurements of a number of people at the establishments at a given time.
  • BRIEF SUMMARY
  • The present disclosure provides for receiving data related to a business, the data including at least historical wait time information, generating a model using one or more computing devices based on at least the historical wait time information for the business, determining an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model, receiving a request for information about a business, and providing for display, in response to the request, the expected wait time. The present disclosure may further include requesting user feedback regarding the expected wait time, and refreshing the model as new wait times and other information are received.
  • Generating the model may include using other types of data in addition to historical wait times, such as place attributes, i.e., particular attributes about a business. Generating the model may also include using historical temporal signals, collected over a look-back window, that fluctuate throughout a day and throughout a week. The historical temporal signals may be further analyzed to derive useful statistics that can also be used in generating the model. The model generated may be a regression model or a classification model, and may further be a linear model, a boosting tree model, a random forest model, or a neural net model.
  • Determining the expected wait time may include determining an upper bound expected wait time. Determining the expected wait time may further include determining whether the expected wait time is less than a predetermined significant-wait threshold. Determining the expected wait time may further include computing a maximum expected wait time for the particular day of the week, and determining at least one peak interval for the particular day of the week having the maximum expected wait time. Determining the expected wait time may also include smoothing the expected wait time data by taking weighted averages with expected wait times of the neighboring time periods.
  • Providing for display may further include displaying the time period as a no-wait time period if the expected wait time is less than the significant threshold. Providing for display may also include displaying the maximum expected wait time of the particular day of the week and the corresponding peak interval. Providing for display may yet include displaying in association with other business information indicating busyness level.
  • The disclosure further provides a system, comprising a memory and one or more processors in communication with the memory. The one or more processors are configured to receive data related to a business, the data including at least historical wait time information, generate a model using one or more computing devices based on at least the historical wait time information for the business, determine an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model, receive a request for information about a business, and provide for display, in response to the request, the expected wait time.
  • The disclosure yet further provides a computer-readable storage medium storing instructions executable by one or more processors for performing a method, comprising receiving data related to a business, the data including at least historical wait time information, generating a model using one or more computing devices based on at least the historical wait time information for the business, determining an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model, receiving a request for information about a business, and providing for display, in response to the request, the expected wait time.
  • BRIEF DESCRIPTION OF THE DRAWINGS
  • FIG. 1 illustrates an example of how expected wait time information may be displayed according to aspects of the disclosure.
  • FIG. 2 is a block diagram illustrating an example system for predicting wait times according to aspects of the disclosure.
  • FIG. 3 is a flow diagram illustrating an example method of generating a wait time prediction model, determining expected wait times using the model, and providing the expected wait times for display in accordance with aspects of the disclosure.
  • FIG. 4 is a flow diagram illustrating an example method of generating a wait time prediction model by training computing devices to learn a pattern from various training data in accordance with aspects of the disclosure.
  • FIG. 5 is another flow diagram illustrating an example of optimizing the predicted wait time for display according to aspects of the disclosure.
  • FIG. 6 is another flow diagram illustrating an example method for determining a maximum expected wait time providing the maximum expected wait time for display according to aspects of the disclosure.
  • FIG. 7 illustrates example displays of expected wait time information according to aspects of the disclosure.
  • DETAILED DESCRIPTION Overview
  • The technology relates generally to predicting and displaying wait times likely to be experienced by users visiting a business or a public service establishment. The technology assists users in making everyday decisions, for example, which grocery store to stop by in the next 30 min; which restaurants nearby have a short wait time for dinner around 7 pm tomorrow; when would be the best time to try that very popular restaurant in town to avoid a very long wait?
  • The expected wait time information may be provided for display. For example, map information related to the business is provided, along with other information corresponding to the business. Examples of such other information include a phone number, directions, website, operating hours, etc. Example of other related information may include a graph showing how popular the business is at various times of a day on various days of a week, relative to the typical peak popularity for the business for the week (“popular times”). The current expected wait time may be provided for display. In addition, the expected wait time for each time period may be overlaid with the popular times graph. The maximum expected wait time for a particular day may be provided for display, along with peak intervals that correspond to the maximum wait time.
  • The wait time prediction technology described herein builds a model based on historical wait times collected from users. As more new wait times are collected, the model may be refined or refreshed. This allows the model to be fine-tuned over time. For example, as more data points are collected, the impact of outliers will be reduced. As another example, new data points can also show certain changes in the wait time patterns for a business (e.g., maybe a place is gaining popularity).
  • The wait time prediction technology may additionally use other types of data in building and refining the model, such as place attributes, i.e., particular attributes about a business. For example, wait time patterns are likely very different for a restaurant as compared to a hospital on a weekend, for a grocery store as compared to a department store on a weekday, or for a fine dining restaurant as compared to a fast food restaurant during lunch hours. Therefore, place category may be play an important role in the wait time pattern. Another important place attribute is whether the business accepts or recommends making a reservation. Yet another place attribute that may play a role in the wait time pattern is the popularity of the business, for example, as reflected by user ratings and comments. Other place attributes, such as online search and visit statistics, may also play a role in the wait time patterns and therefore used in building the prediction model.
  • In addition to the general attributes about the place exemplified above, the wait time prediction technology may further use historical temporal signals that fluctuate throughout a day and throughout a week to build and refine the prediction model. For example, visit data such as the number of arrivals, the number of departures, the occupancy level, and the duration of visits at a particular time period and a particular day of the week can be collected for a predetermined look-back window and used to predict patterns. Other examples include online visits and searches that fluctuate depending on the time of the day and the day of the week.
  • The historical temporal signals described above may be further analyzed over a part of or the entire look-back window to derive useful statistics that can further help predicting wait times at a business, such derived temporal signals may also be used to build and refine the wait time prediction model. For example, a maximum, a minimum, a mean, and a sum of the number of arrivals at a particular time period on a particular day of a week for a business may be computed over half of the look-back window. As another example, aggregated data such as the 90th percentile visit duration at a particular time period on a particular day of a week for a business can be computed over the entire look-back window.
  • As yet another example, lag variables of each of the historical temporal signals and derived temporal signals described above can be determined. The lag variable may be backward-looking or forward-looking. For example, one interesting backward-looking lag variable may be the number of arrivals in the time period preceding the time period for which an expected wait time is to be predicted, since these recent arrivals are likely to remain for more than one time period. As another example, one interesting forward-looking lag variable may be the occupancy level in the upcoming time period following the time period for which an expected wait time is to be predicted. For instance, if the following hour corresponds to the end of lunch break for many employees in the area, a large number of departures may be expected in the upcoming hour, and therefore the business will soon be able to absorb more customers in the upcoming hour.
  • Various types of models may be generated. The model may be, for example, a regression model or a classification model. The model may be a linear model, a boosting tree model, a random forest model, or a neural net model. In one example, a quantile regression model is used to determine an upper bound wait time for the given time period. For example, the 90th percentile expected wait time. A deep neural network with a custom loss function may be used to implement the quantile regression model.
  • The expected wait time may be determined and expressed as an upper bound wait time. For example, “up to 30 min wait” is displayed to the user. The expected wait time may be rounded to a predetermined granularity for display, e.g., instead of “14 min wait” and “36 min wait,” the users will see “15 min wait” and “30 min wait.” In one example, before the expected wait time is displayed to the user, the wait time is smoothed by taking a weighted average of the expected wait time and at least one other expected wait time from a neighboring time period.
  • In many cases, a certain threshold of wait time is almost always expected for a category of businesses. For example, a person walking into almost any sit-down restaurant could expect at least a trivial wait to be seated, say, 10 min. Thus, it would not be interesting for a user to see such trivial wait times. Therefore, in one example, if for a given time period, the expected wait time is less than a predetermined significant-wait threshold, then “usually no wait” may be displayed for the given time period.
  • In another example, the wait time prediction technology determines and displays a maximum expected wait time for the day. In addition, the peak interval including all the time periods that correspond to the maximum expected wait time may be displayed. For example, “longest wait time today⋅60 min⋅12:30 pm-2 pm, 5:00 pm-7:30 pm.” If the expected wait time is less than the predetermined significant-wait threshold for all time periods of a day, then “usually no wait” may be displayed instead of the maximum expected wait time.
  • In another example, feedback is requested from the same user who requested the wait time information. The user can, for example, provide feedback about the accuracy of the expected wait time, input a new wait time actually experienced by the user, or provide additional information about the business or about the user's visit. Such information may be used to refine and refresh the wait time prediction technology.
  • The wait time prediction technology described herein is advantageous in a number of ways. For example, because a larger amount of historical data can be collected as compared to requesting users currently visiting a business for real-time data, the predicted wait times will be less affected by outliers and more indicative of a trend. Further, because the wait time prediction model can be refined and refreshed with newly collected data, the wait time prediction technology can be updated to reflect a recent change in trend. The data used to build the wait time prediction model may be large in volume, but the users are presented with useful information that is simple to understand. Therefore, after a quick glance, users will be aided in their decision making.
  • Example Systems
  • FIG. 1 illustrates an example of how the expected wait time information may be displayed. In this example, graphic 100 includes a map section 110, an informational section 120, and a popular times section 130. The graphic 100 may be displayed, for example on a client device, in response to a request for information related to Business A. The request for information may include, for example, an address, business name, general geographical area, type of business, etc.
  • The map section 110 may include a depiction of geographical objects at a particular geographic location. For example, the geographic objects may include roads, buildings, landmarks, statues, street signs, etc. The objects may be depicted in, for example, a roadgraph, aerial imagery, street level imagery, or the like. The map section 110 may also depict a marker 112 corresponding to a requested location, such as Business A. The marker identifies where the requested location is located with respect to other objects depicted in the map section. The marker may be depicted in any of a variety of shapes and forms. Further, a label 114 may identify that the marker 112 corresponds to Business A. Though not shown, in some examples the marker or map near the marker may be depicted so as to indicate a current expected wait time at Business A. For example, a size, shape, shading, or other aspect of the marker or map may change if the current expected time is short or long. As another example, additional icons may be positioned on the map near the marker to indicate current expected wait time. It should be understood that these are merely examples, and any of a number of indicators may be used.
  • Informational section 120 provides further information related to Business A. Such information may include user input controls 122, 124, 126, enabling the user to take a predetermined action. The predetermined action may be, for example, calling a phone number for Business A, requesting navigational information to Business A, or visiting another website such as a site for Business A. The informational section 120 may also include text 128, such as a listing of the operating hours of Business A.
  • The popular times section 130 includes a graphic representation of how popular the requested location is throughout the day and the week. In this example, a popular times histogram 131 for the current day is shown, the user may select another day 132 to view histograms for other days. In this example, typical levels of busyness are indicated by unshaded bars 133, while the current level of busyness, in real-time, is indicated by a shaded bar 134. In this example, the day is broken down on an hourly basis. The expected wait time 135 for the current hour is displayed as a text above the popular times histogram 131. Below the popular times histogram 131, the maximum expected wait time 136 for the current day, along with the corresponding peak interval, is shown as a text. The user can click on the popular times histogram 131 to see the expected wait time for a different hour, and by selecting a different day to view on the popular times section 130, can view the expected wait times for a different day. Moreover, the timeline 137 may correspond to a range of hours that Business A is in operation.
  • While a number of example sections are described above in connection with FIG. 1, and the expected wait time information is shown as a text, it should be understood that these are merely examples. The expected wait time information may be provided for display in any of a number of ways, such as text, pictorial diagrams, charts, graphs, etc. In other examples, the expected wait time information may be provided to other applications, such as scheduling applications, ride-service applications, games, or any of a variety of other applications.
  • FIG. 2 illustrates an example system used to receive data related to a business, generate a wait time prediction model, determine expected wait times using the generated model, and provide the expected wait times for display upon user request. It should not be considered as limiting the scope of the disclosure or usefulness of the features described herein. In this example, system 200 can include computing devices 210 in communication with one or more computing devices 260, 270, as well as storage system 240, through network 250. Each of the computing devices 210 contains one or more processors 220, memory 230 and other components typically present in general purpose computing devices. Memory 230 of each of the computing devices 210 can store information accessible by the one or more processors 220, including instructions 234 that can be executed by the one or more processors 220.
  • Memory 230 can also include data 232 that can be retrieved, manipulated or stored by the processor. The memory can be of any non-transitory type capable of storing information accessible by the processor, such as a hard-drive, memory card, ROM, RAM, DVD, CD-ROM, write-capable, and read-only memories.
  • The instructions 234 can be any set of instructions to be executed directly, such as machine code, or indirectly, such as scripts, by the one or more processors. In that regard, the terms “instructions,” “application,” “steps,” and “programs” can be used interchangeably herein. The instructions can be stored in object code format for direct processing by a processor, or in any other computing device language including scripts or collections of independent source code modules that are interpreted on demand or compiled in advance. Functions, methods, and routines of the instructions are explained in more detail below.
  • Data 232 may be retrieved, stored, or modified by the one or more processors 220 in accordance with the instructions 234. For instance, although the subject matter described herein is not limited by any particular data structure, the data can be stored in computer registers, in a relational database as a table having many different fields and records, or XML documents. The data can also be formatted in any computing device-readable format such as, but not limited to, binary values, ASCII or Unicode. Moreover, the data can comprise any information sufficient to identify the relevant information, such as numbers, descriptive text, propriety codes, pointers, references to data stored in other memories such as at other network locations, or information that is used by a function to calculate the relevant data.
  • The one or more processors 220 can be any conventional processors, such as a commercially available CPU. Alternatively, the processors can be dedicated components such as an application specific integrated circuit (“ASIC”) or other hardware-based processor. Although not necessary, one or more of the computing devices 210 may include specialized hardware components to perform specific computing processes, such as decoding video, matching video frames with images, distorting videos, encoding distorted videos, etc. faster or more efficiently.
  • Although FIG. 2 functionally illustrates the processor, memory, and other elements of computing devices 210 as being within the same block, the processor, computer, computing device, or memory can actually comprise multiple processors, computers, computing devices, or memories that may or may not be stored within the same physical housing. For example, the memory can be a hard drive or other storage media located in housings different from that of the computing devices 210. Accordingly, references to a processor, computer, computing device, or memory will be understood to include references to a collection of processors, computers, computing devices, or memories that may or may not operate in parallel. For example, the computing devices 210 may include server computing devices operating as a load-balanced server farm, distributed system, etc. Yet further, although some functions described below are indicated as taking place on a single computing device having a single processor, various aspects of the subject matter described herein can be implemented by a plurality of computing devices, for example, communicating information over network 250.
  • Each of the computing devices 210, 260, 270 can be at different nodes of a network 250 and capable of directly and indirectly communicating with other nodes of network 250. Although only a few computing devices are depicted in FIG. 2, it should be appreciated that a typical system can include a large number of connected computing devices, with each different computing device being at a different node of the network 250. The network 250 and intervening nodes described herein can be interconnected using various protocols and systems, such that the network can be part of the Internet, World Wide Web, specific intranets, wide area networks, or local networks. The network can utilize standard communication protocols, such as Ethernet, WiFi and HTTP, protocols that are proprietary to one or more companies, and various combinations of the foregoing. Although certain advantages are obtained when information is transmitted or received as noted above, other aspects of the subject matter described herein are not limited to any particular manner of transmission of information.
  • As an example, each of the computing devices 210 may include web servers capable of communicating with storage system 240 as well as computing devices 260, 270 via the network 250. For example, one or more of server computing devices 210 may use network 250 to transmit and present information to a user on a display, such as display 265 of computing device 260. In this regard, computing devices 260, 270 may be considered client computing devices and may perform all or some of the features described herein.
  • Each of the client computing devices 260, 270 may be configured similarly to the server computing devices 210, with one or more processors, memory and instructions as described above. Each of the client computing devices 260, 270 may be a personal computing device intended for use by a user, and have all of the components normally used in connection with a personal computing device such as a central processing unit (CPU) 262, memory (e.g., RAM and internal hard drives) storing data 263 and instructions 264, a display such as display 265 (e.g., a monitor having a screen, a touch-screen, a projector, a television, or other device that is operable to display information), and user input device 266 (e.g., a mouse, keyboard, touch-screen, or microphone). The client computing device may also include a camera 267 for recording video streams and/or capturing images, speakers, a network interface device, and all of the components used for connecting these elements to one another. The client computing device 260 may also include a location determination system, such as a GPS 268. Other examples of location determination systems may determine location based on wireless access signal strength, images of geographic objects such as landmarks, semantic indicators such as light or noise level, etc.
  • Although the client computing devices 260, 270 may each comprise a full-sized personal computing device, they may alternatively comprise mobile computing devices capable of wirelessly exchanging data with a server over a network such as the Internet. By way of example only, client computing device 260 may be a mobile phone or a device such as a wireless-enabled PDA, a tablet PC, a netbook, a smart watch, a head-mounted computing system, or any other device that is capable of obtaining information via the Internet. As an example the user may input information using a small keyboard, a keypad, microphone, using visual signals with a camera, or a touch screen.
  • As with memory 230, storage system 240 can be of any type of computerized storage capable of storing information accessible by the server computing devices 210, such as a hard-drive, memory card, ROM, RAM, DVD, CD-ROM, write-capable, and read-only memories. In addition, storage system 240 may include a distributed storage system where data is stored on a plurality of different storage devices which may be physically located at the same or different geographic locations. Storage system 240 may be connected to the computing devices via the network 250 as shown in FIG. 1 and/or may be directly connected to any of the computing devices 210.
  • Storage system 240 may store data, such as historical wait time information for particular businesses, place attributes, historical temporal signals, thresholds, etc. For example, the historical wait time information for a popular business for a past number of months or years may be stored.
  • Using the stored data, the computing devices 210 may generate a wait time prediction model for a business and use the model to determine expected wait times at various time periods of a day throughout a week for the business. The computing devices 210 may also receive requests from users' client computing devices 260, 270 about the business and in response, provide for display on the client computing devices 260, 270 the expected wait time at the business.
  • Example Methods
  • Further to example systems described above, example methods are now described. Such methods may be performed using the systems described above, modifications thereof, or any of a variety of systems having different configurations. It should be understood that the operations involved in the following methods need not be performed in the precise order described. Rather, various operations may be handled in a different order or simultaneously, and operations may be added or omitted.
  • FIG. 3 illustrates an example method 300 of receiving data related to a business, generating a wait time prediction model, determining expected wait times using the model, and providing the expected wait times for display in response to a request.
  • In block 310, data related to a business is received. The data may be collected in a number of ways, for example, from databases, from the businesses themselves, from user reports, from real-time measurements, etc. Example data may include the historical wait times experienced at the business at various times of the day, and on different days of the week. The data may also include place attributes, i.e., information about the business, such as the category of the business, reservation information, whether there is usually a wait, online visit statistics, and user ratings and reviews, etc. The data may further include historical temporal signals such as the number of arrivals, number of departures, occupancy level, and duration of visit at different times. The data may also include signals that are derived from such historical temporal signals, for example, a maximum, a minimum, a mean, a sum, a percentile, etc. The data may yet further include lag variables of the historical temporal signals and derived temporal signals. Examples of the data received are described in more detail in FIG. 4.
  • In block 320, a model is generated by training computing devices to learn patterns from the received data. For example, a deep neural network with a custom loss function may be trained with the received data to generate a quantile regression model by minimizing the loss function. For example, the neural network may have two fully connected hidden layers, each with tens or hundreds of neurons. Any of a variety of different types of models may be used, for example, the model may be a regression model or a classification model. The model may also be a linear model, a boosting tree model, a random forest model, or a neural net model. One example of generating a model is described in more detail in FIG. 4.
  • In block 330, the expected wait time is determined using the model generated in block 310. For example, when various place attributes and historical temporal signals about a business without known historical wait times are given to the generated model, expected wait times may be computed for the business using the model.
  • The determining may also include finding an upper bound expected wait time. The determining may further include determining whether the expected wait time is below a predetermined significant-wait threshold. The determining may also include computing the maximum expected wait time for the day. The determining may yet further include smoothing the expected wait time using weighted averages. One example of determining the expected wait time is described in more detail in FIG. 5 and FIG. 6.
  • In block 340, a request is received from a user for information on the business. For example, an example request could be an online search for a particular business, a map search for a particular business, a general search for a category of businesses nearby, or the like.
  • In block 350, the expected wait time is provided for display to the requesting user. The providing for display step may include, for example, providing for display the wait time as an upper bound, providing for display a time period as “no wait” if the expected wait time is below a significant-wait threshold, providing for display the maximum wait time of the day and the corresponding peak intervals, providing for display feedback options, and providing for display expected wait time information in association with other busyness information. The expected wait time information may be provided for display in any of a number of ways, such as text, pictorial diagrams, charts, graphs, etc. In other examples, the expected wait time information may be provided to other applications, such as scheduling applications, ride-service applications, games, or any of a variety of other applications. Some example displays are described in more detail in FIG. 7.
  • In block 360, feedback is requested from the user who requested information on the business. A request for feedback may include, for example, confirmation of whether the displayed expected wait times are consistent with the user's experienced wait time. One example of requesting for feedback is described in more detail in FIG. 7.
  • In block 370, the model is updated using new information received from user feedback. Such new information may be newly experienced wait times, new reservation information, new arrival count at a certain hour, etc. The new information may be used to refresh the training data for generating the model. One example of updating the model is described in more detail in FIG. 4.
  • FIG. 4 illustrates in more detail an example method 400 for generating a wait time prediction model. For example, some of the steps in the method 400 may correspond to blocks 310, 320, and 370.
  • Block 410, example features for generating the model are collected and stored. The data may be collected in a number of ways, for example, from databases, from the businesses themselves, from user reports, etc. Block 420 shows that metadata may be an example feature data that is collected and stored. For example, the metadata could include the time period of the day and the day of the week. In this example, a week is divided into 648 equal time periods, each 30 minutes in length, e.g., Monday 8 am-8:30 am, Monday 8:30 am-9 am, . . . , Sunday 9:30 pm-10 pm, etc. Additionally, the metadata may also include the amount of time since the first opening hour of the business, and the amount of time until the last opening hour of the business.
  • Block 430 shows that place attributes about the business may also be an example feature data that is collected and stored. The data may be collected in a number of ways, for example, from databases, from the businesses themselves, from user reports, etc. Some example place attributes may include the category of the business, number of online visits, reservation information, user review information, etc. The categories could be general, e.g., restaurant, store, hospital, government, etc.; or they could be more nuanced, e.g., fine-dining, fast food, bar, coffee shop, grocery store, department store, clinic, urgent care, post office, DMV, etc. The place attributes may also include online visit statistics, such as number of times an ad about the business is shown as a result of a search, number of online searches about the business, number of online visits of the business's website, number of map searches, etc. The place attributes may additionally include reservation information, such as whether the place accepts reservations, whether it recommends reservations at certain times. User review and rating information may also be place attributes, such as the number of reviewers, the number of comments, the average rating, the number of ratings, etc. Yet another example place attribute is whether the business is known to usually have a wait.
  • Block 440 shows that historical temporal signals may be another example feature data that is collected and stored. Historical temporal signals are signals that may fluctuate throughout a day and throughout a week, therefore, they are collected for each (p, t) tuple, where p is the identifier of the place or the business, and t is the time period for which an expected wait time is to be determined. The look-back window represents the number of data points collected for each (p, t) tuple. For example, if the look-back window is 24 weeks, then there would be 24 data points for each (p, t) tuple. In other words, this means that for Business A at the 8 pm-8:30 pm time period, there are 24 arrival counts recorded over these 24 weeks. The look-back window can be tailored such that a representative amount of historical temporal signals are available, yet not look so far back such that there may have been significant changes in the wait time pattern for the business.
  • Example historical temporal signals include an arrival count, a departure count, an occupancy level, and durations of visits that started in the time period t. The duration of visit data may be the mean duration, the 10th percentile duration, the 90th percentile duration, or the like. Historical temporal signals may also include online behavioral signals, such as the number of times an ad about the business is shown as a result of a search, the number of online searches about the business, the number of online visits of the business's website, the number of map searches, etc.
  • Block 442 shows that derived temporal signals may be yet another example feature data that is collected and stored. For example, a maximum, a minimum, a mean, and a sum of the number of arrivals at a particular time period on a particular day of a week for a business, i.e., a (p, t) tuple, can be computed over the entire look-back window. The same set of statistics can be computed for each of the historical temporal signals described above in relation to block 440. As another example, aggregated data such as the 10th percentile and 90th percentile visit durations for a (p, t) tuple can be computed over the entire look-back window. Instead of using the entire look-back window, alternatively one may also compute derived temporal signals over only part of the look-back window.
  • Block 444 shows that lag variables, yet another example of derived temporal signals, may also be a feature data that is collected and stored. Lag variables may be determined for each of the historical temporal and derived temporal signals shown in blocks 440 and 442. A lag variable may be backward-looking or forward-looking. For example, one interesting backward-looking lag variable may be the number of arrivals in an earlier time period of the day, such as in the time period t−1 preceding the time period t for which an expected wait time is to be predicted, since these recent arrivals are likely to remain beyond the time period t. As another example, one interesting forward-looking lag variable may be the occupancy level in a later time period of the day, such as in the upcoming time period t+1 following the time period t for which an expected wait time is to be predicted. For instance, if the following hour corresponds to the end of lunch break for many employees in the area, a large number of departures may be expected in the upcoming hour, and therefore the business will soon be able to absorb more customers in the upcoming hour.
  • In block 450, example labels for generating the model are collected and stored. Block 460 shows that historical wait time data may be an example label data that is collected and stored. The historical wait time data may be collected from consumer surveys. For example, the survey may ask a user to input the actual wait time experienced, or it may ask a user to choose from a list of estimates, e.g., 0 min, 5 min, 10 min, . . . , 30 min, 60 min, more than 60 min, etc. The survey may also ask the user at what time and day the wait time was experienced. The survey may additionally ask the user whether service was eventually received at the business. Additionally, the survey may ask more nuanced questions such as whether a reservation was made, was it dine in or take out, etc. To ensure that the received wait time data are accurate, the surveys are preferably sent to the user soon after the visit. The method is not limited to using surveys, for example, the actual wait time data may be collected from another database, from the businesses, or from real-time measurements, etc.
  • The historical wait time data collected are sorted and stored for each (p, t) tuple. For example, if a wait time is collected for a visit to Business A at 5:13 pm on Monday, it will be sorted as a wait time for Business A, Monday 5 pm-5:30 pm.
  • In block 470, a deep neural network with a custom loss function is trained with the example features collected and stored in block 410 and the example labels collected and stored in block 450. For example, the neural network may have two fully connected hidden layers, each with tens or hundreds of neurons. In this example, the deep neural network learns patterns from the feature data and label data and generates a quantile regression model by minimizing the loss function. In this example, the quantile regression model predicts an expected wait time as a value at a predetermined quantile, specifically, the 90th percentile expected wait time.
  • However, the method 400 is not limited to training a neural network to generate a quantile regression model. Various types of models may be generated using the method 400. For example, the model may be a regression model or a classification model. The model may also be a linear model, a boosting tree model, a random forest model, or a neural net model, etc.
  • In block 480, the training data are refreshed with feedback collected from users. For example, the user who requested the wait time information may provide feedback about the accuracy of the expected wait time, or input a new wait time actually experienced by the user. The feedback may include additional options such as leave a review or a rating, or provide more information about the business. Feedback may additionally be collected via other methods and from other sources other than the requesting user, for example, they could be collected from new customer surveys, databases, etc.
  • For example, the feedback may include a newly experienced actual wait time, which becomes an additional set of label data that can be used to refine and update the current model. The feedback may also include new information that a user provides about the business, for example, that the place recommends reservations for lunch. Therefore, the current model may be updated with the new place attribute about reservations. The feedback may also include a new duration of visit that the user reports, which can be used to update the historical temporal signals. The model may be updated each time a new feedback is collected, or alternatively and more practically, for example, at some regular time interval, whenever a predetermined number of new feedbacks are collected, or whenever a new place attribute needs to be updated, etc.
  • FIG. 5 illustrates in more detail an example method 500 for determining an expected wait time using the model generated from the method 400 and providing for display. For example, some of the steps in the method 500 may correspond to blocks 330 and 350.
  • Block 510 shows that an upper bound expected wait time is determined using the model generated from method 400. An upper bound expected wait time may be more useful for decision making than, for example, a lower bound or the mean expected wait time, since a user may have limited availability, e.g., lunch hour ends. Instead of the absolute upper bound, one preferred example determines the 90th percentile expected wait time. Alternatively, the upper bound could be the 70th percentile expected wait time, and so on. The method is not limited to expressing the expected wait time as an upper bound, for example, the expected wait time may also be expressed as a range of typically expected wait times or the most likely expected wait time.
  • In block 520, it is determined whether the expected wait time is for a time period outside of opening hours, and if so, the expected wait time is reset as 0 min (block 530).
  • In block 540, the wait time is smoothed by taking a weighted average of the expected wait time and at least one other expected wait time from a neighboring time period. For example, if the expected wait time for Friday 5 pm-5:30 pm is 20 min, for 5:30 pm-6 pm is 55 min, and for 6 pm-6:30 pm is 70 min, then an example weighted average for the 5:30 pm-6 pm time period may be (0.5×20 min+1×55 min+0.5×70 min)/2=50 min. Expected wait times outside of the opening hours are not used in calculating weighted averages.
  • In block 550, the expected wait time is rounded to a predetermined granularity. For example, if granularity is set at 15 min, “up to 30 min wait” and “up to 45 min wait” may be provided for display instead of “up to 28 min wait” and “up to 40 min wait.” Continuing from the example above for the time period Friday 5:30 pm-6 pm, the expected wait time of 50 min may be rounded to 45 min.
  • In block 560, the expected wait time is compared to a significant-wait threshold. In many cases, a certain threshold of wait time is almost always expected for a category of businesses. For example, a person walking into almost any sit-down restaurant without a reservation could expect at least some wait before seated, say, 10 min. Thus, it would not be interesting for a user to see such trivial wait times. Since upper bound expected wait times are determined in this example, a significant-wait threshold of 30 min may be preferable. Therefore, continuing from the examples above, for Friday 5 pm-5:30 pm the expected wait time is less than 30 min, so “usually no wait” will be provided for display (block 570) for Friday 5 pm-5:30 pm. For Friday 5:30 pm-6 pm in the example above, the expected wait time is greater than 30 min, therefore “up to 45 min wait” will be provided for display (block 580) for Friday 5:30 pm-6 pm.
  • FIG. 6 illustrates in more detail an example method 600 for determining a maximum expected wait time using the model generated from the example method 400 and providing the maximum expected wait time for display. For example, some of the steps in the method 600 may correspond to blocks 330 and 350.
  • In block 610, a maximum expected wait time for the day is determined among all the expected wait times available for the day.
  • In block 620, it is determined whether the maximum expected wait time of the day is less than the predetermined significant-wait threshold. If so, “usually no wait” is provided for display for the day 630. For example, if the significant-wait threshold is set at 30 min and the maximum wait time of the day is 15 min, “usually no wait” would be provided for display; otherwise, the method carries on to determine the peak interval corresponding to the maximum expected wait time. In this example, the threshold for determining whether there is a significant wait for a particular time period is the same as the threshold for determining whether there is a significant wait for a day; however, alternatively, two different thresholds can be set.
  • In block 640, the peak interval corresponding to the maximum expected wait time is determined. In this example, it is the longest interval including all the time periods that have the maximum expected wait time for the day. For example, if Friday from 7 pm-8:30 pm, the expected wait time is at maximum for the day at 90 min, then 7 pm-8:30 pm would be the peak interval for Friday, which includes all three time periods 7 pm-7:30 pm, 7:30 pm-8 pm, and 8 pm-8:30 pm. The method may determine multiple peak intervals for a given day.
  • In block 650, it is determined whether the peak interval is only 30 min in length. If the peak interval is greater than 30 min in length, as in the example above, the maximum expected wait time along with the corresponding peak interval will be provided for display (block 660). If the peak interval is only 30 min in length, it is determined whether any of its neighbors have an expected wait time that is only one granularity less than the maximum expected wait time (block 670). If so, such neighbors will be included in the peak interval (block 680) and the expected wait time of the entire peak interval will be reset to the majority value of the peak interval (block 690). For example, if Thursday 7 pm-7:30 pm has the maximum expected wait time of 45 min and it is the longest interval for this maximum expected wait time, then the peak interval is only 30 min in length. However, the expected wait time is 30 min for both 6:30 pm-7 pm and at 7:30 pm-8 pm on Thursday. Since the granularity is set at 15 min, the expected wait times for both 6:30 pm-7 pm and 7:30 pm-8 pm are only one granularity below the maximum expected wait time of 45 min. Therefore, the new peak interval will be 6:30 pm-8 pm on Thursday, and the expected wait time for this new peak interval will be the majority value of 30 min. In this way, instead of showing a very short peak interval of 30 min for Thursday having an expected wait of 45 min, it may be better to show the user a broader peak interval that may be more representative of the peak, since the two time periods around 7 pm-7:30 pm are not much different.
  • FIG. 7 illustrates example displays 700 of the expected wait time data. In this example, the current expected wait time data is displayed along with the popular times data, including real-time busyness data. In this example, a user sent a request for information about the business at 7:12 pm on a Wednesday, therefore the expected wait time 710 for the time period 7 pm-7:30 pm on a Wednesday is shown. The expected wait time 710 is expressed as an upper bound, “up to 45 min wait.” The text “Correct?” links the requesting user to a feedback form, where the user may provide feedback about whether the information is accurate, an actually experienced wait time, additional information about the business, or other information related to the user's visit.
  • Pop-up 720 shows that, when the user clicks on the bar representing the time period 8 pm-9 pm on Wednesday, the higher expected wait time between the two time periods 8 pm-8:30 pm and 8:30 pm-9 pm is displayed. In this example, the expected wait time for 8 pm-8:30 pm is 15 min and the expected wait time for 8:30 pm-9 pm is 0 min. Therefore, the expected wait time for the time period 8 pm-9 pm on Wednesday is expressed as “usually no wait” since both values are below the predetermined significant-wait threshold. In this example, the threshold is set at 30 minutes.
  • Below the popular times histogram, a maximum expected wait time 730 for the current day is displayed along with the corresponding peak time interval. In this example, the maximum expected wait time 730 for Wednesday is displayed as “longest wait today⋅45 min⋅5:30 pm-7 pm.” The text “Useful?” links the user to a feedback form, where the user may provide feedback about whether the information is accurate, provide an actually experienced wait time, additional information about the business, or other information related to the user's visit.
  • Pop-up 740 shows that, when the user changes the date on the dropdown menu for dates 750 to Monday, the expected wait time information for Monday is displayed. In this example, Monday's maximum expected wait time 760 is expressed as “usually no wait” because there is no time period on Monday with an expected wait time equal or above the predetermined significant-wait threshold, which is set at 30 minutes in this example.
  • Although not shown here, if the business accepts or recommends making reservations at particular times or on particular days, and the expected wait times during these times or days are below the predetermined significant-wait threshold, it may simply be that most of the visitors make reservations to the business during such times and days, and therefore that was the reason for usually no wait. In such cases, though not shown in FIG. 7, the method may provide for display “recommend reservations” instead of “usually no wait,” as the latter may be misleading.
  • It should be understood that the examples of how the predicted wait times are displayed are not limiting, and that the information may be displayed in any of a variety of other formats. For example, the expected wait time information may be provided for display as text, pictorial diagrams, charts, graphs, etc. In other examples, the expected wait time information may be provided to other applications, such as scheduling applications, ride-service applications, games, or any of a variety of other applications.
  • The wait time prediction technology described herein is advantageous in a number of ways. For example, the technology may be developed for a large scale of businesses and public service establishments, and therefore can be relied on by users in making many everyday decisions. Further, because a larger amount of historical data may be collected as compared to requesting users currently visiting a business for real-time data, the predicted wait times will be less affected by outliers and more indicative of a trend. Still further, because the wait time prediction model can be refined and refreshed with newly collected data, the wait time prediction technology can be updated to reflect a recent change in trend. The data used to build the wait time prediction model may be large in volume, but the users are presented with useful information that is simple to understand. Therefore, after a quick glance, users will be aided in their decision making.
  • Unless otherwise stated, the foregoing alternative examples are not mutually exclusive, but may be implemented in various combinations to achieve unique advantages. As these and other variations and combinations of the features discussed above can be utilized without departing from the subject matter defined by the claims, the foregoing description of the examples should be taken by way of illustration rather than by way of limitation of the subject matter defined by the claims. In addition, the provision of the examples described herein, as well as clauses phrased as “such as,” “including” and the like, should not be interpreted as limiting the subject matter of the claims to the specific examples; rather, the examples are intended to illustrate only one of many possible examples. Further, the same reference numbers in different drawings can identify the same or similar elements.

Claims (20)

1. A method, comprising:
receiving data related to a business, the data including at least historical wait time information;
generating a model using one or more computing devices based on at least the historical wait time information for the business;
determining an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model;
receiving a request for information about a business; and
providing for display, in response to the request, the expected wait time.
2. The method according to claim 1, wherein generating the model is additionally based on place attributes, the place attributes comprise at least one of a place category, a number of online visits, reservation information, and user review information.
3. The method according to claim 1, wherein generating the model is additionally based on historical temporal signals collected for a predetermined look-back window for the particular time period on the particular day of the week, the historical temporal signals comprise at least one of an arrival count, a departure count, an occupancy level, durations of visits, and a number of online visits.
4. The method according to claim 3, wherein the historical temporal signals further comprise derived temporal signals computed over at least a part of the predetermined look-back window for at least one of the historical temporal signals for the particular time period on the particular day of the week.
5. The method according to claim 4, wherein the derived temporal signals further comprise at least one of a backward-looking lag variable, the backward-looking lag variable being one of the historical temporal signals collected for an earlier time period on the particular day of the week.
6. The method according to claim 4, wherein the derived temporal signals further comprise at least one of a forward-looking lag variable, the forward-looking lag variable being one of the historical temporal signals collected for a later time period on the particular day of the week.
7. The method according to claim 1, wherein the expected wait time is an upper bound expected wait time.
8. The method according to claim 1, further comprising:
determining whether the expected wait time is less than a predetermined significant-wait threshold; and
displaying the particular time period as a no-wait time period if the expected wait time is less than the predetermined significant-wait threshold.
9. The method according to claim 1, further comprising:
computing a maximum expected wait time for the particular day of the week;
determining at least one peak interval for the particular day of the week, the peak interval comprising the particular time periods having the maximum expected wait time;
displaying the maximum expected wait time; and
displaying the at least one peak interval.
10. The method according to claim 1, wherein the expected wait time is smoothed by taking a weighted average of the expected wait time and at least one other expected wait time from a neighboring time period.
11. The method according to claim 1, wherein the providing for display is in association with other business information indicating busyness level.
12. The method according to claim 1, further comprising:
requesting user feedback regarding the expected wait time;
receiving, in response to the request, the user feedback; and
updating the model based on the user feedback.
13. The method according to claim 1, further comprising:
refreshing the model as new wait times are received.
14. The method according to claim 1, wherein the model is one of a regression model or a classification model.
15. The method according to claim 14, wherein the model is a quantile regression model capable of determining the expected wait time as a value at a predetermined quantile.
16. The method according to claim 1, wherein the model is one of a linear model, a boosting tree model, a random forest model, or a neural net model.
17. A system, comprising:
a memory; and
one or more processor in communication with the memory, the one or more processors configured to:
receive data related to a business, the data including at least historical wait time information;
generate a model using one or more computing devices based on at least the historical wait time information for the business;
determine an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model;
receive a request for information about a business; and
provide for display, in response to the request, the expected wait time.
18. The system according to claim 17, wherein generating the model is additionally based on place attributes, the place attributes comprise at least one of a place category, a number of online visits, reservation information, and user review information.
19. The system according to claim 17, wherein generating the model is additionally based on historical temporal signals collected for a predetermined look-back window for the particular time period of the day on the particular day of the week, the historical temporal signals comprise at least one of an arrival count, a departure count, an occupancy level, durations of visits, and a number of online visits.
20. A computer-readable storage medium storing instructions executable by one or more processors for performing a method, comprising:
receiving data related to a business, the data including at least historical wait time information;
generating a model using one or more computing devices based on at least the historical wait time information for the business;
determining an expected wait time for service at the business for at least one particular time period on a particular day of a week using the model;
receiving a request for information about a business; and
providing for display, in response to the request, the expected wait time.
US15/800,787 2017-11-01 2017-11-01 Wait time prediction Pending US20190130322A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/800,787 US20190130322A1 (en) 2017-11-01 2017-11-01 Wait time prediction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
US15/800,787 US20190130322A1 (en) 2017-11-01 2017-11-01 Wait time prediction

Publications (1)

Publication Number Publication Date
US20190130322A1 true US20190130322A1 (en) 2019-05-02

Family

ID=66243856

Family Applications (1)

Application Number Title Priority Date Filing Date
US15/800,787 Pending US20190130322A1 (en) 2017-11-01 2017-11-01 Wait time prediction

Country Status (1)

Country Link
US (1) US20190130322A1 (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110750716A (en) * 2019-09-27 2020-02-04 恒大智慧科技有限公司 Automatic recommendation method of scenic spots in intelligent scenic spot, computer equipment and storage medium
US10673784B1 (en) * 2016-08-25 2020-06-02 C/Hca, Inc. Processing delay predictions based on queue assessments
CN113553609A (en) * 2021-09-17 2021-10-26 支付宝(杭州)信息技术有限公司 Method and system for predicting service by combining multiple parties based on privacy protection
US20220207992A1 (en) * 2020-12-30 2022-06-30 Here Global B.V. Surprise pedestrian density and flow
US11455669B2 (en) * 2018-01-30 2022-09-27 Walmart Apollo, Llc Method and apparatus for predicting wait times
US20230401269A1 (en) * 2021-12-30 2023-12-14 Google Llc System And Method For Representing Busyness Using Content Specific To A Place

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140180848A1 (en) * 2012-12-20 2014-06-26 Wal-Mart Stores, Inc. Estimating Point Of Sale Wait Times
US20150081348A1 (en) * 2013-09-13 2015-03-19 Scott C. Avera Systems and methods using crowd sourced wait time information
US20150088671A1 (en) * 2013-09-20 2015-03-26 Sony Corporation Providing customized entertainment based on wait time
US20150088562A1 (en) * 2013-09-24 2015-03-26 Celia Maria Woods Restaurant selection, wait time and attendance management
US20170046800A1 (en) * 2015-08-10 2017-02-16 Google Inc. Systems and Methods of Automatically Estimating Restaurant Wait Times Using Wearable Devices

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140180848A1 (en) * 2012-12-20 2014-06-26 Wal-Mart Stores, Inc. Estimating Point Of Sale Wait Times
US20150081348A1 (en) * 2013-09-13 2015-03-19 Scott C. Avera Systems and methods using crowd sourced wait time information
US20150088671A1 (en) * 2013-09-20 2015-03-26 Sony Corporation Providing customized entertainment based on wait time
US20150088562A1 (en) * 2013-09-24 2015-03-26 Celia Maria Woods Restaurant selection, wait time and attendance management
US20170046800A1 (en) * 2015-08-10 2017-02-16 Google Inc. Systems and Methods of Automatically Estimating Restaurant Wait Times Using Wearable Devices

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10673784B1 (en) * 2016-08-25 2020-06-02 C/Hca, Inc. Processing delay predictions based on queue assessments
US11455669B2 (en) * 2018-01-30 2022-09-27 Walmart Apollo, Llc Method and apparatus for predicting wait times
CN110750716A (en) * 2019-09-27 2020-02-04 恒大智慧科技有限公司 Automatic recommendation method of scenic spots in intelligent scenic spot, computer equipment and storage medium
US20220207992A1 (en) * 2020-12-30 2022-06-30 Here Global B.V. Surprise pedestrian density and flow
CN113553609A (en) * 2021-09-17 2021-10-26 支付宝(杭州)信息技术有限公司 Method and system for predicting service by combining multiple parties based on privacy protection
US20230401269A1 (en) * 2021-12-30 2023-12-14 Google Llc System And Method For Representing Busyness Using Content Specific To A Place

Similar Documents

Publication Publication Date Title
US20190130322A1 (en) Wait time prediction
US10318967B2 (en) End-to-end effective citizen engagement via advanced analytics and sensor-based personal assistant capability (EECEASPA)
Miličević et al. An investigation of the relationship between destination branding and destination competitiveness
US10567568B2 (en) User event pattern prediction and presentation
RU2721176C2 (en) Systems and methods for predicting user behavior based on location data
KR101512278B1 (en) Using mobile device data to create a storyline, model user routine and personality, and create customized recommendation agents
EP2325794A1 (en) Generating an activity inference model from contextual data
JP2019508766A (en) System, method, and device for generating a heat map of a geographical area
Huang et al. Predicting human mobility with activity changes
WO2016191515A1 (en) Personalized information from venues of interest
US11361421B1 (en) Visual attractiveness scoring system
US9785897B2 (en) Methods and systems for optimizing efficiency of a workforce management system
US11200449B1 (en) Image ranking system
US20140179354A1 (en) Determining contact opportunities
US20210150484A1 (en) Machine-learning creation of job posting content
Lansley et al. Challenges to representing the population from new forms of consumer data
Supak et al. An open source web-mapping system for tourism planning and marketing
JP5798983B2 (en) Place evaluation system, apparatus, method and program
Magano et al. Mobile Apps and Travel Apps on the tourism journey
Hladík et al. Spatial-temporal analysis of retail and services using Facebook Places data: a case study in Brno, Czech Republic
US11138615B1 (en) Location-based place attribute prediction
US20230186189A1 (en) Method and system for intelligently managing facilities
Boo et al. Tourists’ online reviews of convention centers
Hu et al. Task-oriented information value measurement based on space-time prisms
CN105378626A (en) Situation-aware presentation of information

Legal Events

Date Code Title Description
AS Assignment

Owner name: GOOGLE LLC, CALIFORNIA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:CHEN, LI-SUE;CHIEN, STEVE;DUONG, QUANG;AND OTHERS;SIGNING DATES FROM 20171102 TO 20171212;REEL/FRAME:044372/0174

STPP Information on status: patent application and granting procedure in general

Free format text: DOCKETED NEW CASE - READY FOR EXAMINATION

STPP Information on status: patent application and granting procedure in general

Free format text: NON FINAL ACTION MAILED

STPP Information on status: patent application and granting procedure in general

Free format text: RESPONSE TO NON-FINAL OFFICE ACTION ENTERED AND FORWARDED TO EXAMINER

STPP Information on status: patent application and granting procedure in general

Free format text: FINAL REJECTION MAILED

STPP Information on status: patent application and granting procedure in general

Free format text: FINAL REJECTION MAILED

STCV Information on status: appeal procedure

Free format text: NOTICE OF APPEAL FILED

STCV Information on status: appeal procedure

Free format text: APPEAL BRIEF (OR SUPPLEMENTAL BRIEF) ENTERED AND FORWARDED TO EXAMINER

STCV Information on status: appeal procedure

Free format text: EXAMINER'S ANSWER TO APPEAL BRIEF MAILED

STCV Information on status: appeal procedure

Free format text: ON APPEAL -- AWAITING DECISION BY THE BOARD OF APPEALS

STCV Information on status: appeal procedure

Free format text: BOARD OF APPEALS DECISION RENDERED

STCV Information on status: appeal procedure

Free format text: APPLICATION INVOLVED IN COURT PROCEEDINGS

STCV Information on status: appeal procedure

Free format text: COURT PROCEEDINGS TERMINATED