CN115860790A - Method for predicting place passenger flow, method and device for training model - Google Patents
Method for predicting place passenger flow, method and device for training model Download PDFInfo
- Publication number
- CN115860790A CN115860790A CN202211415102.0A CN202211415102A CN115860790A CN 115860790 A CN115860790 A CN 115860790A CN 202211415102 A CN202211415102 A CN 202211415102A CN 115860790 A CN115860790 A CN 115860790A
- Authority
- CN
- China
- Prior art keywords
- passenger flow
- passenger
- prediction model
- time point
- training
- 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
Links
Images
Abstract
The embodiment of the application provides a method for predicting the passenger flow of a place, a method and a device for training a model, wherein the method for predicting the passenger flow of the place comprises the following steps: acquiring passenger flow influence factors of a place at a target time point and a plurality of historical passenger flows of a plurality of historical time points before the target time point; inputting a plurality of historical passenger flows serving as time sequence data into a passenger flow prediction model trained in advance to obtain an initial predicted passenger flow of a target time point; determining a passenger flow correction coefficient based on the passenger flow influence factor; and correcting the initial predicted passenger flow based on the passenger flow correction coefficient to obtain the target predicted passenger flow at the target time point. According to the method and the device, the change rule of the passenger flow along with time and the influence factors of the passenger flow are comprehensively considered, and the accuracy of the passenger flow prediction result of the place can be effectively improved.
Description
Technical Field
The application relates to the technical field of data processing, in particular to a place passenger flow prediction method, a model training method and a device.
Background
At present, many places have the demand of passenger flow volume prediction, and for example, shopping malls are taken as examples, and can determine reasonable holding time for propaganda or promotion activities by predicting passenger flow volume, thereby improving the activity effect. In the related art, the passenger flow volume is usually directly predicted according to passenger flow influence factors such as holidays, temperatures and the like, and the problem that the passenger flow volume prediction is not accurate enough exists.
Disclosure of Invention
The embodiment of the application provides a method for predicting passenger flow in a place, a method and a device for training a model, and aims to solve the problem that the passenger flow prediction is not accurate enough in the related technology.
In order to solve the technical problem, the present application is implemented as follows:
in a first aspect, an embodiment of the present application provides a method for predicting a place passenger flow, including:
acquiring passenger flow influence factors of a place at a target time point and a plurality of historical passenger flows of a plurality of historical time points before the target time point;
inputting a plurality of historical passenger flows serving as time sequence data into a passenger flow prediction model trained in advance to obtain an initial predicted passenger flow of a target time point;
determining a passenger flow correction coefficient based on the passenger flow influence factor;
and correcting the initial predicted passenger flow based on the passenger flow correction coefficient to obtain the target predicted passenger flow at the target time point.
In a second aspect, an embodiment of the present application further provides a model training method, including:
acquiring a training sample, wherein the training sample comprises a first passenger flow of a marked time point, a plurality of second passenger flows of a plurality of historical time points before the marked time point and passenger flow influence factors of the marked time point;
training a passenger flow prediction model based on a training sample, and obtaining the trained passenger flow prediction model under the condition that the loss value of a loss function in the passenger flow prediction model meets a preset condition; the passenger flow prediction model takes a plurality of second passenger flows as input and outputs a third passenger flow, and the loss function calculates a loss value according to the third passenger flow and the first passenger flow;
inputting a plurality of second passenger flows into the trained passenger flow prediction model to obtain a fourth passenger flow;
and obtaining a trained influence factor prediction model based on the passenger flow influence factor of the marked time point, the fourth passenger flow and the first passenger flow training influence factor prediction model.
In a third aspect, an embodiment of the present application further provides a device for predicting a passenger flow in a place, including:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring passenger flow influence factors of a place at a target time point and a plurality of historical passenger flows of a plurality of historical time points before the target time point;
the input module is used for inputting a plurality of historical passenger flows serving as time sequence data into a passenger flow prediction model trained in advance to obtain initial predicted passenger flows of a target time point;
the determining module is used for determining a passenger flow correction coefficient based on the passenger flow influence factor;
and the correction module is used for correcting the initial predicted passenger flow based on the passenger flow correction coefficient to obtain the target predicted passenger flow at the target time point.
In a fourth aspect, an embodiment of the present application further provides a model training apparatus, including:
the fourth acquisition module is used for acquiring a training sample, wherein the training sample comprises a first passenger flow volume of a marked time point, a plurality of second passenger flow volumes of a plurality of historical time points before the marked time point and passenger flow volume influence factors of the marked time point;
the first training module is used for training a passenger flow prediction model based on training samples, and obtaining a passenger flow prediction model after training under the condition that the loss value of a loss function in the passenger flow prediction model meets a preset condition; the passenger flow prediction model takes a plurality of second passenger flows as input and outputs a third passenger flow, and the loss function calculates a loss value according to the third passenger flow and the first passenger flow;
the prediction module is used for inputting the plurality of second passenger flows into the passenger flow prediction model after training to obtain a fourth passenger flow;
and the second training module is used for obtaining a trained influence factor prediction model based on the passenger flow influence factor of the marked time point, the fourth passenger flow and the first passenger flow training influence factor prediction model.
In a fifth aspect, an embodiment of the present application further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the method described above is implemented.
In a sixth aspect, the present application further provides a computer-readable storage medium, where a computer program is stored, and when executed by a processor, the computer program implements the method described above.
According to the method for predicting the passenger flow of the place, the passenger flow influence factor of the place at the target time point and a plurality of historical passenger flows of a plurality of historical time points before the target time point are obtained; inputting a plurality of historical passenger flows serving as time sequence data into a passenger flow prediction model trained in advance to obtain an initial predicted passenger flow at a target time point; determining a passenger flow correction coefficient based on the passenger flow influence factor; and correcting the initial predicted passenger flow based on the passenger flow correction coefficient to obtain the target predicted passenger flow at the target time point. When the method and the device for predicting the passenger flow of the place predict the passenger flow of the place, the rule of the passenger flow changing along with time and the influence factors of the passenger flow are comprehensively considered, and the accuracy of the passenger flow prediction result of the place can be effectively improved.
Drawings
Fig. 1 is a schematic flow chart of a method for predicting passenger flow in a place according to an embodiment of the present disclosure;
FIG. 2 is a schematic flow chart illustrating a model training method according to an embodiment of the present disclosure;
fig. 3 is a schematic structural diagram of a device for predicting passenger flow in a place according to an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram of a model training device according to an embodiment of the present application.
Detailed Description
To make the technical problems, technical solutions and advantages to be solved by the present application clearer, the following detailed description is made with reference to the accompanying drawings and specific embodiments. In the following description, specific details such as specific configurations and components are provided only to help the embodiments of the present application be fully understood. Accordingly, it will be apparent to those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. In addition, descriptions of well-known functions and constructions are omitted for clarity and conciseness.
Unless otherwise defined, technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this application belongs. As used in this application, the terms "first," "second," and the like do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. Also, the use of the terms "a" or "an" and the like do not denote a limitation of quantity, but rather denote the presence of at least one.
As shown in fig. 1, a method for predicting place passenger flow provided in an embodiment of the present application includes:
103, determining a passenger flow correction coefficient based on the passenger flow influence factor;
and 104, correcting the initial predicted passenger flow based on the passenger flow correction coefficient to obtain the target predicted passenger flow at the target time point.
In the embodiment of the present application, the place may be a shopping mall, a tourist location, a transportation hub (for example, a bus stop, a train station), or a public transportation (for example, a bus or a subway), and for simplifying the description, the method for predicting the passenger flow in the place provided in the embodiment of the present application will be mainly described below by taking the place as a shopping mall as an example.
The main body of execution of the method for predicting the passenger flow in the place is an electronic device, such as a mobile terminal, a personal computer, a server, or the like, and is not limited herein.
In step 101, the electronic device may obtain a passenger volume influencing factor of the place at a target time point and a plurality of historical passenger volumes at a plurality of historical time points before the target time point.
For example, the target time point may be 1/15/2022, and the plurality of historical time points may be 14 days, i.e., 1/2022 to 14/1/2022. Alternatively, the target time point is 17 pm of a certain day, and the plurality of historical time points may be time points corresponding to 7 hours from 10 am to 16 pm of the day.
It is easy to understand that there may be a historical passenger flow corresponding to each historical time point, and the historical passenger flow may be obtained by the prior art. For example, a camera can be arranged in a shopping mall, and the passenger flow in unit time is counted through a machine vision technology; for another example, in a subway, the passenger flow per unit time can be counted by counting the number of people passing through the gate. The historical passenger flow of each historical time point can be obtained by counting the passenger flow in unit time.
The passenger flow influencing factors can be preset according to needs, and in some examples, the passenger flow influencing factors are weather, holidays, geographical positions of places, competitive goods around the places, activities of the places, vehicles entering and exiting the places and the like.
In step 102, the electronic device may input a plurality of historical passenger flows as time series data into a passenger flow prediction model trained in advance, so as to obtain an initial predicted passenger flow at a target time point.
The passenger flow prediction model may be a prediction model built based on a neural network, or may also be an Artificial Intelligence (AI) model of another type, such as a support vector machine, and the like, which is not limited herein.
It is easy to understand that the data format of the input and output of the passenger flow prediction model may be preset, and in this embodiment, the data format of the input of the passenger flow prediction model is time series data, such as the passenger flow (corresponding to the historical passenger flow) of each day of the continuous preset number of days. The data format of the output of the traffic prediction model may then be a single value, such as the traffic for a certain day (corresponding to the initial predicted traffic); of course, the output of the passenger flow prediction model may also be a probability value of a preset classification result, the probability value may be in a positive correlation or a negative correlation with the passenger flow, and the passenger flow can be predicted indirectly based on the probability value.
The passenger flow prediction model used in this embodiment is pre-trained, and the format of the training sample is matched with the input and output of the passenger flow prediction model in step 102, for example, the training sample includes time series data and a label of the time series data, and the label may correspond to the actual passenger flow. The specific training process of the passenger flow prediction model is not described in detail here.
In step 103, the electronic device may determine a passenger volume correction coefficient based on the passenger volume influencing factor, and in step 104, the electronic device may further correct the initial predicted passenger volume using the passenger volume correction coefficient to obtain a target predicted passenger volume at the target time point.
As indicated above, the passenger flow influencing factors may include items such as weather, holidays, geographical location of the venue, and the like. In one embodiment, the item value-correction value correspondence may be set for each item, for example, the item value "sunny day, temperature 22 to 25 ° corresponds to a correction value of +5%, and the item value" typhoon "corresponds to a correction value of-40%. Therefore, when the weather forecast shows that the weather is 'clear day, temperature is 22-25 degrees' at the target time point, the target predicted passenger flow can be obtained by multiplying 1.05 on the basis of the initial predicted passenger flow; when the weather forecast shows that the weather is typhoon at the target time point, the target predicted passenger flow volume can be obtained by multiplying 0.6 on the basis of the initial predicted passenger flow volume.
Of course, in other embodiments, the item values of the items of the passenger flow influencing factors may be input into a predetermined regression model or a pre-trained AI model to obtain the target predicted passenger flow.
In this embodiment, the initial predicted passenger flow may be a passenger flow predicted based on a change rule of the place passenger flow with time, and the target predicted passenger flow is obtained by further correcting the initial predicted passenger flow on the basis of considering a passenger flow influence factor, so that the target predicted passenger flow comprehensively considers the change rule of the passenger flow with time and the passenger flow influence factor.
The target predicted passenger volume may be used for various purposes. Taking a place as a market as an example, the target predicted passenger flow can provide reference for the activities held by market practitioners; taking a transportation junction as an example, the target predicted passenger flow can provide reference for dispatching of traffic control personnel and the like.
According to the method for predicting the passenger flow of the place, the passenger flow influence factor of the place at the target time point and a plurality of historical passenger flows of a plurality of historical time points before the target time point are obtained; inputting a plurality of historical passenger flows serving as time sequence data into a passenger flow prediction model trained in advance to obtain an initial predicted passenger flow of a target time point; determining a passenger flow correction coefficient based on the passenger flow influence factor; and correcting the initial predicted passenger flow based on the passenger flow correction coefficient to obtain the target predicted passenger flow at the target time point. When the method and the device for predicting the passenger flow of the place predict the passenger flow of the place, the rule of the passenger flow changing along with time and the influence factors of the passenger flow are comprehensively considered, and the accuracy of the passenger flow prediction result of the place can be effectively improved.
In some embodiments, the passenger flow affecting factors include at least one of: weather, holidays, geographical location of the venue, arena offerings around the venue, venue activities, and vehicle ingress and egress to the venue.
Taking the passenger flow influencing factor of weather as an example, generally speaking, the passenger flow of a shopping mall can be positively influenced in sunny days and at proper temperature and humidity, and the passenger flow of the shopping mall can be negatively influenced in typhoon, rainy days, hot days or cold days.
Similarly, for the traffic influencing factor of holidays, when the target time point is holiday, the traffic of the shopping mall can be positively influenced. Of course, it is readily understood that holidays can affect the type of traffic in addition to the traffic volume. For example, children are on a regular basis, family-type guests are increased, while lovers are increased when love is in the middle of the day.
The geographical location of the venue can also be a factor in passenger traffic. For example, given the geographic location of a mall, the population density of the city of the mall, the city class (e.g., first-line city, second-line city), the surrounding environment (e.g., number of residences and business office buildings, number of residential grades, etc.), the surrounding traffic (e.g., number of subway stations and instances, number of bus stations and shifts), and the type and quality of the surrounding population may be derived from relevant information channels. In general, the population density and the surrounding traffic directly affect the traffic volume of the shopping mall, and the factors such as the city level and the surrounding environment may indirectly affect the traffic volume of the shopping mall from the perspective of the consumption capacity.
The competitive products around the market may be other shopping centers, and the information of the number, the scale, the state contact ratio and the like of the shopping centers can influence the passenger flow of the market.
The venue activity may be an activity of a mall or mall business organization, which typically has a positive impact on traffic when venue activity is present.
The vehicles at the entrance and exit places can be identified through visual equipment (such as a camera with an image processor) at the entrance and exit gate of the parking lot, and the identified information can comprise the number, the model and the like of the vehicles. The number of vehicles can directly influence the passenger flow of a shopping mall; and can reversely deduce shopping mall customers and the like according to the vehicle model.
The passenger flow volume influence factors can be obtained in a corresponding mode, for example, information such as weather, holidays and the like can be obtained from a related server, data such as competitive products around a place, place activities and the like can be manually input and pre-recorded in electronic equipment, vehicles entering and exiting the place can be directly obtained from the visual equipment, and the like.
Optionally, step 103, determining a passenger flow correction coefficient based on the passenger flow influence factor, includes:
and inputting the passenger flow influence factors into an influence factor prediction model to obtain passenger flow correction coefficients, wherein the influence factor prediction model is a regression model or a pre-trained classification model.
As indicated above, the passenger flow influencing factors may include items such as weather, holidays, geographical location of the venue, and the like. In one embodiment, the electronic device may perform regression fitting on each item value described above in advance to obtain a regression model. The regression model may correspond to an impact factor prediction model for determining a passenger flow correction factor based on the input passenger flow impact factor.
The regression model may be a linear regression model or a nonlinear regression model, and is not particularly limited herein. Taking a linear regression model as an example, the following linear relationship can be established:
K＝αA+βB+γC+…
k is a passenger flow correction coefficient A, B, C, \8230, alpha, beta, gamma, \8230isan item value of each item in the passenger flow influence factors, and is a regression coefficient.
Alpha, beta, gamma and 8230are obtained through linear fitting, and by combining an example, after the passenger flow prediction model is trained, the passenger flow at a time point can be predicted by using the passenger flow prediction model, the actual passenger flow at the time point can be obtained through monitoring, a passenger flow correction coefficient can be determined based on the two passenger flows, and the passenger flow correction coefficient and the passenger flow influence factor at the time point can be used as a group of data for calculating alpha, beta, gamma and 8230.
In yet another embodiment, the passenger flow correction factor may also be determined based on the passenger flow influencing factors by using a classification model, such as a neural network, a support vector machine, or a random forest. As for the acquisition manner of the training samples of the classification model, it may be similar to the acquisition manner of the data set used for the above regression coefficient calculation, and a repetitive description will not be made here. The classification model is fully trained through the training samples, and the influence factor prediction model can be obtained.
In one example, the classification model may be a multi-classification model, which may output classification results such as "-20%", "-15%", \ 8230; "+20%", etc., which may be directly used as the passenger flow volume correction coefficients. In another example, the classification model may be a binary classification model, and the classification result and the corresponding probability value may be output, where the probability value is positively or negatively correlated with the passenger flow correction coefficient, and the passenger flow correction coefficient may be indirectly obtained based on the probability value and the correlations.
Optionally, after obtaining the target predicted passenger flow volume at the target time point, the method further includes:
acquiring a crowd image of a place;
and generating recommendation information based on the target predicted passenger flow volume, the crowd portrayal and a preset recommendation rule, wherein the preset recommendation rule comprises the corresponding relation among the passenger flow volume, the crowd portrayal and the recommendation information.
The crowd portrait can be a portrait established for a place passenger group, and by taking a market scene as an example, the crowd portrait can be obtained by analyzing shopping behaviors of a plurality of customers. In some examples, a crowd representation may be described as, for example, a family, a young woman, a co-worker, a backpack guest, a white-collar, a family of cars, and so forth.
In some embodiments, the crowd images may be pre-stored in the electronic device, or the electronic device may receive the crowd images from other servers, which is not limited herein.
The electronic equipment generates recommendation information according to the target predicted passenger flow volume, the crowd portrait and a preset recommendation rule.
In combination with some application scenarios, the recommendation information may be mall positioning recommendation information, mall preparation-period recruiting recommendation information, operation-period recruiting recommendation information, brand and mall operation recommendation information, or activity planning adjustment recommendation information.
Taking the mall positioning recommendation information as an example, it may include crowd oriented (e.g., young white collar, family, etc.), business state (e.g., high end, luxury, public, etc.), functional area size (e.g., shopping area, parking lot area, etc.), architectural design (e.g., subway, bus, cell design recommendation, etc.), and so on. In general, the electronic device may perform preliminary positioning on a shopping mall from different angles by predicting the passenger flow and drawing the crowd of the passenger group, so as to generate the relevant recommendation information.
As for other types of recommendation information, the market preparation period recruitment recommendation information can be used for a market practitioner to perform brand recruitment on the basis of the amount of customers and different customers; the operation period recruiter recommendation information can be used for the market practitioner to perform brand adjustment (settlement of poor brands) in a targeted manner in combination with the passenger flow in the shop; the brand and the market operation recommendation information can be used for carrying out brand operation adjustment (position adjustment and brand joint marketing activities) by market practitioners in combination with the passenger flow and the brand turnover in the market; the activity plan adjustment recommendation information can be used for targeted activity plan (brand joint marketing activity, holiday activity and other activities) of market practitioners.
The recommendation information may be automatically generated by the electronic device according to a preset recommendation rule. In some embodiment modes, the preset recommendation rule may be a table recording correspondence between the passenger flow volume, the people portrait and the recommendation information.
For example, the preset recommendation rules may include a correspondence between "passenger volume >5000 people/day" (corresponding to passenger volume), "young white matter percentage >40%" (corresponding to people figures), and "holding a luxury brand promotion campaign" (corresponding to recommendation information).
As another example, the preset recommendation rules may include a correspondence between "passenger volume >6000 people/day" (corresponding to passenger volume), "family proportion >30%" (corresponding to people figures), and "introduction of interest training institutions" (corresponding to recommendation information).
Of course, in some possible embodiments, the preset recommendation rule may also be implemented by an algorithm such as a fuzzy algorithm.
In the embodiment, the target predicted passenger flow volume and the crowd portrait are used for generating the recommendation information, so that the rationality of the recommendation information can be effectively guaranteed, and the application value of the recommendation information is improved.
Optionally, in a case that the place is a shopping mall, before obtaining the crowd image of the place, the method further includes:
acquiring business data of a market and a character label constructed for characters in the market;
and generating a crowd image of the place according to the business data and the person label.
In some examples, the business data for a mall may include the sales of various types of merchants, the number of customers, and so on. The overall customer group category of the shopping mall can be determined according to the business data of the shopping mall, for example, when the business value of an interest training institution or a parent-child amusement park is higher, it is indicated that the customer group category of the shopping mall may be mainly home; when the light luxury class merchant has a high turnover, it is stated that the class of customers in the mall may be white-collar dominated, and so on.
In other examples, a store practitioner may investigate people in a customer base by on-line or off-line, and build a people label based on the information of the investigation. Such as randomly selecting a person in a store for inquiry by way of interview, or organizing a code scanning activity to allow each person to participate in an online evaluation, etc.
The questions of the survey may include "age", "frequency of shopping mall", "primary consumption content", "consumption amount zone", and the like, and according to these questions, a character tag may be generated for the character, such as "entertainment", "eating", "shopping", and the like.
The business data and the data such as the person tag may be manually counted and then input into the electronic device, or the electronic device may directly communicate with the relevant terminal to obtain the business data and the data, such as the person tag, or the data may be obtained by processing the obtained raw data, which is not described in detail herein.
The electronic equipment generates a crowd image of the place according to the business data and the person label. In conjunction with one example, the crowd representation may include proportion information for a family, young women, co-workers, backpackers, white-collar workers, family of vehicles, and the like. First proportion information of various crowds can be obtained according to business data, second proportion information of various crowds can be obtained according to the person labels, and crowd images of places can be obtained through weighted averaging of the first proportion information and the second proportion information. Of course, the manner of generating the portrait image is described here by way of example, and the specific manner of generating the portrait image in actual use may be adjusted as necessary.
According to the embodiment, the crowd portrayal of the shopping mall can be generated accurately according to the business data and the person label, and the reasonability of the recommendation information obtained based on the crowd portrayal is improved.
As shown in fig. 2, an embodiment of the present application further provides a model training method, including:
and 204, obtaining a trained influence factor prediction model based on the passenger flow influence factor of the marked time point, the fourth passenger flow and the first passenger flow training influence factor prediction model.
In the model training method of the embodiment of the application, the training process of the passenger flow prediction model and the influence factor prediction model is included, and in general, based on the training samples, the passenger flow prediction model can be trained first, after the passenger flow prediction model is trained, the model parameters of the passenger flow prediction model are fixed, and then the influence factor prediction model is trained further.
To facilitate understanding of the specific process of model training in this embodiment, the first passenger flow volume labeled at a time point may be defined as FT T And marking a plurality of second passenger flow volumes of a plurality of historical time points before the time point as { FT 1 ，FT 2 ，FT 3 ，…，FT N And recording the passenger flow influence factors of the marked time points as INF T Wherein, the subscript N is the number of the historical time points, and T represents the labeled time points.
Will { FT 1 ，FT 2 ，FT 3 ，…，FT N Inputting the passenger flow rate into an initial passenger flow rate prediction model (which can be understood as a passenger flow rate prediction model which is not fully trained), outputting a third passenger flow rate, and recording the third passenger flow rate as FT F With loss function in the passenger flow prediction model, according to FT T And FT F A loss value of the loss function may be determined, and model parameters of the passenger flow prediction model may be adjusted based on the loss value such that the loss value as a whole tends to decrease.
After the passenger flow prediction model is fully trained based on the training samples, the loss value of the loss function can be smaller than the preset value, and based on the loss value of the loss function smaller than the preset value, the loss value of the loss function can be used as a preset condition for representing the passenger flow prediction model to obtain full training. Of course, in practical applications, the loss value of the loss function meeting the preset condition may also be set according to actual needs, for example, the loss value is sequentially decreased in a preset training period, and the like.
After training of the passenger flow prediction model is completed, { FT ] may be set 1 ，FT 2 ，FT 3 ，…，FT N Inputting the fourth passenger flow volume into a passenger flow volume prediction model after training to obtain a fourth passenger flow volume which is marked as FT F ′。
INF can be used when training the influencing factor prediction model T As input to the influence factor prediction model and based on the influence factorOutput of the measurement model and (FT) T /FT F ' -1) adjusting relevant parameters, such as regression coefficients or network parameters, in the influence factor prediction model, so as to obtain the trained influence factor prediction model.
Certainly, the above is some exemplary descriptions of the training processes of the passenger flow prediction model and the influence factor prediction model, in practical applications, the training samples used by the two models may be the same or different, and the basic concept of the training processes can be ensured to be realized.
In the embodiment, the passenger flow prediction model and the influence factor prediction model are trained, so that the passenger flow prediction model can predict the passenger flow according to the change rule of the passenger flow along with time, and the influence factor prediction model can correct the passenger flow output by the passenger flow prediction model based on the passenger flow influence factor, thereby improving the accuracy of the passenger flow obtained by final prediction.
In some embodiments, the influence factor prediction model is a regression model or a classification model, which may be selected according to actual needs.
As shown in fig. 3, an embodiment of the present application further provides a device for predicting passenger flow in a place, including:
a first obtaining module 301, configured to obtain a passenger flow influence factor of a place at a target time point and a plurality of historical passenger flows at a plurality of historical time points before the target time point;
the input module 302 is configured to input a plurality of historical passenger flows serving as time sequence data to a passenger flow prediction model trained in advance, so as to obtain an initial predicted passenger flow at a target time point;
a determining module 303, configured to determine a passenger flow correction coefficient based on the passenger flow influence factor;
and the correction module 304 is configured to correct the initial predicted passenger volume based on the passenger volume correction coefficient to obtain a target predicted passenger volume at the target time point.
Optionally, the determining module 303 is specifically configured to:
and inputting the passenger flow influence factors into an influence factor prediction model to obtain passenger flow correction coefficients, wherein the influence factor prediction model is a regression model or a pre-trained classification model.
Optionally, the location passenger flow prediction device may further include:
the second acquisition module is used for acquiring crowd images of the places;
the first generation module is used for generating recommendation information based on the target predicted passenger flow volume, the crowd portrayal and a preset recommendation rule, wherein the preset recommendation rule comprises the corresponding relation among the passenger flow volume, the crowd portrayal and the recommendation information.
Optionally, the location passenger flow prediction device may further include:
the third acquisition module is used for acquiring business data of the market and a character label constructed for characters in the market under the condition that the market is the market;
and the second generation module is used for generating the crowd image of the place according to the business data and the person label.
Optionally, the passenger flow influencing factors include at least one of: weather, holidays, geographical location of the venue, arena offerings around the venue, venue activities, and vehicle ingress and egress to the venue.
The device for predicting the passenger flow in the place provided by the embodiment of the application is a device right corresponding to the method for predicting the passenger flow in the place of the embodiment, and the method embodiment can be applied to the device embodiment and obtain the same technical effect, and the details are not repeated here.
As shown in fig. 4, an embodiment of the present application further provides a model training apparatus, including:
a fourth obtaining module 401, configured to obtain a training sample, where the training sample includes a first passenger flow volume at a time point, a plurality of second passenger flow volumes at a plurality of historical time points before the time point, and a passenger flow volume influence factor at the time point;
the first training module 402 is used for training a passenger flow prediction model based on a training sample, and obtaining the passenger flow prediction model after training under the condition that the loss value of a loss function in the passenger flow prediction model meets a preset condition; the passenger flow prediction model takes a plurality of second passenger flows as input and outputs a third passenger flow, and the loss function calculates a loss value according to the third passenger flow and the first passenger flow;
the prediction module 403 is configured to input the multiple second passenger flows into the passenger flow prediction model after training to obtain a fourth passenger flow;
the second training module 404 is configured to obtain a trained impact factor prediction model based on the passenger flow impact factor at the marked time point, the fourth passenger flow, and the first passenger flow training impact factor prediction model.
Optionally, the impact factor prediction model is a regression model or a classification model.
The model training device provided in the embodiment of the present application is a device authority corresponding to the model training method in the above embodiment, and the method embodiment may be applied to the device embodiment to achieve the same technical effect, which is not described herein again.
The embodiment of the present application further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the aforementioned place passenger flow prediction method or model training method is implemented.
The embodiment of the application also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the method for predicting the passenger flow in the place or the method for training the model are implemented.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules, so as to perform all or part of the functions described above. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working processes of the units and modules in the system may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to the related descriptions of other embodiments for parts that are not described or illustrated in a certain embodiment.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the technical solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus/terminal device and method may be implemented in other ways. For example, the above-described embodiments of the apparatus/terminal device are merely illustrative, and for example, a module or a unit may be divided into only one logical function, and may be implemented in other ways, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be in an electrical, mechanical or other form.
Units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow in the method of the embodiments described above can be realized by a computer program, which can be stored in a computer readable storage medium and used to instruct related hardware, and when the computer program is executed by a processor, the steps of the method embodiments described above can be realized. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, U.S. disk, removable hard disk, magnetic disk, optical disk, computer Memory, read-Only Memory (ROM), random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution media, and the like.
The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present application and are intended to be included within the scope of the present application.
Claims (10)
1. A method for predicting a venue passenger flow, comprising:
acquiring passenger flow influence factors of a place at a target time point and a plurality of historical passenger flows of a plurality of historical time points before the target time point;
inputting the plurality of historical passenger flows serving as time sequence data into a passenger flow prediction model trained in advance to obtain initial predicted passenger flows of the target time point;
determining a passenger flow correction coefficient based on the passenger flow influence factor;
and correcting the initial predicted passenger flow based on the passenger flow correction coefficient to obtain the target predicted passenger flow at the target time point.
2. The method of claim 1, wherein determining a passenger flow correction factor based on the passenger flow influencing factor comprises:
and inputting the passenger flow influence factors into an influence factor prediction model to obtain the passenger flow correction coefficient, wherein the influence factor prediction model is a regression model or a pre-trained classification model.
3. The method of claim 1, wherein after obtaining the target predicted passenger flow volume for the target time point, the method further comprises:
acquiring a crowd portrait of the place;
and generating recommendation information based on the target predicted passenger flow volume, the crowd portrayal and a preset recommendation rule, wherein the preset recommendation rule comprises the corresponding relation among the passenger flow volume, the crowd portrayal and the recommendation information.
4. The method of claim 3, wherein, in the case where the venue is a mall, prior to said obtaining the representation of the crowd at the venue, the method further comprises:
acquiring business data of a market and a character label constructed for characters in the market;
and generating a crowd portrait of the place according to the business data and the person labels.
5. The method of claim 1, wherein the passenger flow affecting factors comprise at least one of: weather, holidays, geographical location of the venue, arena offerings around the venue, venue activities, and vehicle ingress and egress to the venue.
6. A method of model training, comprising:
acquiring a training sample, wherein the training sample comprises a first passenger flow of a marked time point, a plurality of second passenger flows of a plurality of historical time points before the marked time point, and a passenger flow influence factor of the marked time point;
training a passenger flow prediction model based on the training sample, and obtaining the trained passenger flow prediction model under the condition that the loss value of a loss function in the passenger flow prediction model meets a preset condition; the passenger flow volume prediction model takes the second passenger flow volumes as input and outputs a third passenger flow volume, and the loss function calculates a loss value according to the third passenger flow volume and the first passenger flow volume;
inputting the plurality of second passenger flows into a passenger flow prediction model after training to obtain a fourth passenger flow;
and obtaining a trained influence factor prediction model based on the passenger flow influence factor of the marked time point, the fourth passenger flow and the first passenger flow training influence factor prediction model.
7. The method of claim 6, wherein the influencer prediction model is a regression model or a classification model.
8. A venue passenger flow prediction device, comprising:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring passenger flow influence factors of a place at a target time point and a plurality of historical passenger flows of a plurality of historical time points before the target time point;
the input module is used for inputting the plurality of historical passenger flows serving as time sequence data into a passenger flow prediction model trained in advance to obtain the initial predicted passenger flow of the target time point;
the determining module is used for determining a passenger flow correction coefficient based on the passenger flow influence factor;
and the correction module is used for correcting the initial predicted passenger flow based on the passenger flow correction coefficient to obtain the target predicted passenger flow at the target time point.
9. A model training apparatus, comprising:
the fourth acquisition module is used for acquiring a training sample, wherein the training sample comprises a first passenger flow volume of a marked time point, a plurality of second passenger flow volumes of a plurality of historical time points before the marked time point and passenger flow volume influence factors of the marked time point;
the first training module is used for training a passenger flow prediction model based on the training samples, and obtaining the passenger flow prediction model after training under the condition that the loss value of a loss function in the passenger flow prediction model meets a preset condition; the passenger flow volume prediction model takes the second passenger flow volumes as input and outputs a third passenger flow volume, and the loss function calculates a loss value according to the third passenger flow volume and the first passenger flow volume;
the prediction module is used for inputting the plurality of second passenger flow volumes into a passenger flow volume prediction model after training to obtain a fourth passenger flow volume;
and the second training module is used for obtaining a trained influence factor prediction model based on the passenger flow influence factor of the marked time point, the fourth passenger flow and the first passenger flow training influence factor prediction model.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which, when executed by a processor, implements the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211415102.0A CN115860790A (en) | 2022-11-11 | 2022-11-11 | Method for predicting place passenger flow, method and device for training model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211415102.0A CN115860790A (en) | 2022-11-11 | 2022-11-11 | Method for predicting place passenger flow, method and device for training model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115860790A true CN115860790A (en) | 2023-03-28 |
Family
ID=85663235
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211415102.0A Pending CN115860790A (en) | 2022-11-11 | 2022-11-11 | Method for predicting place passenger flow, method and device for training model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115860790A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117557300A (en) * | 2024-01-12 | 2024-02-13 | 湖南大学 | Method and system for deducing business liveness based on energy consumption data of main equipment |
| CN117575684A (en) * | 2024-01-15 | 2024-02-20 | 杭州路过网络有限公司 | Passenger flow volume prediction method and system |
-
2022
- 2022-11-11 CN CN202211415102.0A patent/CN115860790A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117557300A (en) * | 2024-01-12 | 2024-02-13 | 湖南大学 | Method and system for deducing business liveness based on energy consumption data of main equipment |
| CN117557300B (en) * | 2024-01-12 | 2024-04-05 | 湖南大学 | Method and system for deducing business liveness based on energy consumption data of main equipment |
| CN117575684A (en) * | 2024-01-15 | 2024-02-20 | 杭州路过网络有限公司 | Passenger flow volume prediction method and system |
| CN117575684B (en) * | 2024-01-15 | 2024-04-05 | 杭州路过网络有限公司 | Passenger flow volume prediction method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Suel et al. | Development of joint models for channel, store, and travel mode choice: Grocery shopping in London | |
| Buldeo Rai et al. | How does consumers’ omnichannel shopping behaviour translate into travel and transport impacts? Case-study of a footwear retailer in Belgium | |
| Suel et al. | Incorporating online shopping into travel demand modelling: challenges, progress, and opportunities | |
| Lenk et al. | Hierarchical Bayes conjoint analysis: recovery of partworth heterogeneity from reduced experimental designs | |
| Applebaum | Methods for determining store trade areas, market penetration, and potential sales | |
| Savitt | Historical research in marketing | |
| CN115860790A (en) | Method for predicting place passenger flow, method and device for training model | |
| Rizzi et al. | The impact of traffic images on travel time valuation in stated-preference choice experiments | |
| Inoa et al. | Effect of an accessibility measure in a model for choice of residential location, workplace, and type of employment | |
| CN111539764B (en) | Big data multiple access selection method based on submodular function | |
| Alkhalifah et al. | Enhancement of E-commerce Service by Designing Last Mile Delivery Platform. | |
| Kedia | Estimating the effect of online shopping and collection-delivery points on shopping travel in New Zealand. | |
| Minh et al. | Factors Affecting Customers’ Satisfaction with E-commerce of Business in Vietnam | |
| Nematizadeh et al. | Designing an Online Advertising Model with a GIF Marketing Approach (Case: Iran’s Oil and Gas Industrial Tourism Hubs) | |
| Schifferes | Shopping for shared value | |
| Khangembam | Impact of e-commerce on speciality stores in categorised shopping centres | |
| Roumpani et al. | Data-driven modelling of public library infrastructure and usage in the United Kingdom | |
| Krämer et al. | Automated valuation models: improving model performance by choosing the optimal spatial training level | |
| Subhani et al. | Location-Wise House Prediction Using Data Science Techniques | |
| Liu et al. | Modeling sequences of discrete and continuous variables over time with an application to the vehicle ownership and usage problem | |
| Najafabadi | Designing an On-demand Dynamic Crowdshipping Model and Evaluating Its Ability to Serve Local Retail Delivery in New York City | |
| Al-Makhmari et al. | A Study on Attracting Local And International Investments and Improving an Integrated Logistics System Allows Local Products to Reach Global Markets | |
| Armona et al. | Information Inequality in Online Education | |
| US20220207465A1 (en) | System, Method, and Platform for Determining Optimal Equipment Purchase and Installation Parameters Using Neural Networks | |
| Sierag | Revenue Management in the Hotel Industry: from practice to theory |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |