CN112434872A - Hotel yield prediction method, system, equipment and storage medium - Google Patents

Hotel yield prediction method, system, equipment and storage medium Download PDF

Info

Publication number
CN112434872A
CN112434872A CN202011394090.9A CN202011394090A CN112434872A CN 112434872 A CN112434872 A CN 112434872A CN 202011394090 A CN202011394090 A CN 202011394090A CN 112434872 A CN112434872 A CN 112434872A
Authority
CN
China
Prior art keywords
yield
hotel
data
network model
feature
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
CN202011394090.9A
Other languages
Chinese (zh)
Inventor
林晨
褚煜佳
李鹤
孙刚
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.)
Ctrip Computer Technology Shanghai Co Ltd
Original Assignee
Ctrip Computer Technology Shanghai Co Ltd
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 Ctrip Computer Technology Shanghai Co Ltd filed Critical Ctrip Computer Technology Shanghai Co Ltd
Priority to CN202011394090.9A priority Critical patent/CN112434872A/en
Publication of CN112434872A publication Critical patent/CN112434872A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION 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
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0202—Market predictions or forecasting for commercial activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION 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
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/12—Hotels or restaurants

Abstract

The invention provides a hotel yield prediction method, a hotel yield prediction system, hotel yield prediction equipment and a storage medium, wherein the method comprises the following steps: obtaining yield data of a plurality of hotels, wherein the yield data comprises historical yield data, hotel attribute characteristic data and current booking progress data; constructing a quantile loss function based on the quantile regression model; constructing an initial deep learning network model; training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to respectively obtain a target network model corresponding to each hotel; respectively predicting the yield of a plurality of hotels in a preset time period in the future based on the target network model, and outputting the yield value of each hotel in the preset time period; according to the method and the device, the yield of multiple hotels can be predicted simultaneously, and the accuracy of hotel yield prediction is improved.

Description

Hotel yield prediction method, system, equipment and storage medium
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a hotel yield prediction method, a system, equipment and a storage medium.
Background
In the prior art, a traditional time series model or a neural network model LSTM (Long Short-Term Memory artificial neural network) is generally adopted for predicting the yield of a hotel (namely, the sales volume of resources such as hotel rooms). However, the traditional time sequence model is difficult to comprehensively consider the influence of a plurality of exogenous variables on the final yield, and the expression capacity of the traditional time sequence model is limited. In the hotel yield prediction process, exogenous variables such as a scheduled progress characteristic, a competition circle characteristic of the hotel, a hotel attribute characteristic, a date characteristic and the like are involved besides the time sequence characteristic. The prediction accuracy of the conventional timing model is significantly poor. The neural network model LSTM has limited expression capability in a scene of predicting multiple hotels, and cannot be used in a scene of predicting multiple hotels at the same time.
Disclosure of Invention
Aiming at the problems in the prior art, the invention aims to provide a hotel yield prediction method, a hotel yield prediction system, hotel yield prediction equipment and a hotel yield prediction storage medium, which can realize simultaneous prediction of the yields of multiple hotels and improve the accuracy of hotel yield prediction.
In order to achieve the above purpose, the invention provides a hotel yield prediction method, which comprises the following steps:
obtaining yield data of a plurality of hotels;
constructing a quantile loss function based on the quantile regression model;
constructing an initial deep learning network model;
training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to respectively obtain a target network model corresponding to each hotel;
and respectively predicting the yield of the plurality of hotels in a preset time period in the future based on the target network model, and outputting the yield value of each hotel in the preset time period.
Optionally, the training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to obtain target network models corresponding to the hotels respectively includes:
vectorizing the yield data of each hotel to obtain a plurality of characteristic vectors belonging to the hotel;
respectively performing characteristic low-order crossing and characteristic high-order crossing on the characteristic vector to respectively obtain a first predicted value and a second predicted value;
carrying out weighted summation on the first predicted value and the second predicted value to obtain an initial predicted result, and carrying out normalization operation on the initial predicted result by using a preset activation function to obtain a predicted probability value;
and taking the prediction probability value and a preset real result value as the input of the quantile loss function to obtain a target network model.
Optionally, the initial deep learning network model includes a DNN model, and the step of performing feature low-order intersection and feature high-order intersection on the feature vector to obtain a first predicted value and a second predicted value respectively includes:
optionally combining two feature vectors in all the feature vectors to form a plurality of feature vector sets, wherein each feature vector set comprises two feature vectors;
performing inner product operation on all the characteristic vector sets respectively to obtain a first predicted value;
and taking all the characteristic vectors as the input of the DNN model to obtain a second predicted value.
Optionally, the target network model includes quantiles, and the quantiles in the target network models corresponding to different hotels are different.
Optionally, the training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to obtain target network models corresponding to the hotels respectively includes:
for each hotel, respectively taking a plurality of preset quantiles as the input of the quantile loss function, training to generate a plurality of different second network models belonging to the same hotel, wherein each second network model corresponds to one preset quantile;
based on the yield data of the past N days as a test set, respectively testing the plurality of different second network models, and taking the second network model with the smallest difference between the obtained predicted value and the real yield value as a target network model corresponding to the hotel; the production data for the past N days includes the true production value.
Optionally, the training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to obtain target network models corresponding to the hotels respectively includes:
generating a feature ID corresponding to each feature according to the features contained in the yield data of each hotel;
generating a feature dictionary file according to the features and the feature IDs;
storing the feature dictionary file into the target network model.
Optionally, each of the yield data includes a feature and a feature value corresponding to the feature; the initial deep learning network model has a plurality of parameters; each of the features matches one of the parameters; and each said feature matches one said feature vector;
the step of performing inner product operation on all the feature vector sets respectively to obtain a first predicted value includes: performing inner product operation on each characteristic vector set according to a low-order cross function;
the low order cross function is:
Figure BDA0002813965160000031
wherein the content of the first and second substances,
Figure BDA0002813965160000032
y1representing said first predicted value, w0A first parameter, w, representing said initial deep learning network modeliAn ith parameter representing the initial deep learning network model, m representing a total number of features per one of the production data, xiRepresenting a feature value corresponding to an ith feature in each of the production data,<vi,vj>representing a feature vector viAnd vjInner product of viA feature vector, v, representing the corresponding ith feature in each of said yield datajA feature vector, x, representing the corresponding jth feature in each of said yield datajA characteristic value, v, representing the value of the j-th characteristic in each of said production datai,fRepresenting a feature vector viThe f-th element of (1), vj,fRepresenting a feature vector vjThe f-th element in (1), k denotes a feature vector viTotal number of elements in (1).
Optionally, the quantile loss function is:
Figure BDA0002813965160000033
wherein n represents the total amount of the production data, q represents a preset quantile, y'pA predicted value, y, representing said yield data of the p-th slicepThe actual yield value of the p-th piece of the yield data is represented, and L represents the total loss value.
Optionally, the method further comprises the step of:
updating the production data for the plurality of hotels based on the hotel production data for the current time period;
and training the initial deep learning network model again based on the updated yield data to obtain an updated target network model and the updated quantiles.
Optionally, the production data includes historical production data, hotel attribute feature data, and current booking progress data; the hotel attribute characteristic data comprises hotel star level, house type level and hotel city data, and the yield data further comprises historical booking progress data, current canceling progress, hotel competition circle characteristic and date characteristic.
The invention also provides a hotel yield prediction system, which is used for realizing the hotel yield prediction method and comprises the following steps:
the system comprises a yield data acquisition module, a reservation module and a reservation module, wherein the yield data acquisition module is used for acquiring yield data of a plurality of hotels, and the yield data comprises historical yield data, hotel attribute characteristic data and current reservation progress data;
the loss function building module is used for building a quantile loss function based on the quantile regression model;
the model building module is used for building an initial deep learning network model;
the model training module is used for training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to respectively obtain target network models corresponding to the hotels;
and the yield prediction module is used for predicting the yields of the plurality of hotels in a preset time period in the future based on the target network model and outputting the yield values of the hotels in the preset time period.
The invention also provides hotel yield prediction equipment, which comprises:
a processor;
a memory having stored therein executable instructions of the processor;
wherein the processor is configured to perform the steps of any of the hotel yield prediction methods described above via execution of the executable instructions.
The present invention also provides a computer readable storage medium storing a program which when executed by a processor implements the steps of any of the hotel yield prediction methods described above.
Compared with the prior art, the invention has the following advantages and prominent effects:
according to the hotel yield prediction method, the hotel yield prediction system, the hotel yield prediction equipment and the hotel yield prediction storage medium, the influence of a plurality of exogenous variables on the final yield can be comprehensively considered, and the hotel yield prediction accuracy is improved; the method has the parallel prediction capability for multiple hotels, can generate models by adopting different quantiles according to different hotels, and ensures the accuracy of predicting the yield of each hotel in parallel prediction on the premise of realizing simultaneous prediction of multiple hotels.
Drawings
Other features, objects and advantages of the present invention will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, with reference to the accompanying drawings.
Fig. 1 is a schematic diagram of a hotel yield prediction method according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating the step S40 according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating step S402 according to an embodiment of the present invention;
FIG. 4 is a flowchart illustrating step S404 according to an embodiment of the present invention;
FIG. 5 is a flowchart illustrating step S40 according to another embodiment of the present invention;
fig. 6 is a schematic diagram of a hotel yield prediction method according to another embodiment of the present invention;
fig. 7 is a schematic structural diagram of a hotel yield prediction system disclosed in an embodiment of the present invention;
fig. 8 is a schematic structural diagram of hotel yield prediction equipment disclosed by the embodiment of the invention;
fig. 9 is a schematic structural diagram of a computer-readable storage medium according to an embodiment of the disclosure.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar structures, and thus their repetitive description will be omitted.
As shown in fig. 1, the embodiment of the invention discloses a hotel yield prediction method, which comprises the following steps:
s10, obtaining production data for a plurality of hotels, which may include historical production data, hotel attribute feature data, and current booking progress data. The hotel attribute feature data can comprise hotel star level, house type level and hotel city data, and the yield data further comprises historical booking progress data, current canceling progress, hotel competition circle feature and date feature. The application is not limited to the above-mentioned yield data and hotel attribute feature data, for example, the above-mentioned yield data may further include a hotel comment feature.
S20, constructing a quantile loss function based on the quantile regression model. Each quantile loss function comprises a corresponding preset quantile, each hotel corresponds to one preset quantile, and the preset quantiles in the quantile loss functions of different hotels can be different or the same. The setting can be made as required by those skilled in the art.
In this embodiment, the quantile loss function is:
Figure BDA0002813965160000061
where n represents the total number of the above-mentioned production data, i.e., the total number of the production data. q represents a preset quantile and may be, for example, 0.45, 0.5 or 0.6. y'pPredicted value, y, representing the p-th yield datapThe actual yield value representing the p-th yield data. The yield data comprises the actual yield value corresponding to the data. L represents the total loss.
∑p:y′p≥yp(y′p-yp) Representing that for the p production data, if the predicted value is greater than or equal to the true real production value, namely y'p≥ypThen execute
Figure BDA0002813965160000062
The total loss value was calculated as the loss value of the p-th yield data. Y 'if the predicted value is less than the true production value'p<ypThen execute
Figure BDA0002813965160000063
The total loss value was calculated as the loss value of the p-th yield data.
And S30, constructing an initial deep learning network model. In this embodiment, the initial Deep learning network model includes a Deep Neural Networks (DNN) model. The initial deep learning network model may be a deep learning model of deep fm. This is not limited by the present application. The deep FM model includes two parts, namely FM (factor Machine) and DNN (DNN).
And S40, training the initial deep learning network model based on the quantile loss function and the yield data of the hotels, and respectively obtaining a target network model corresponding to each hotel. Specifically, as shown in fig. 2, step S40 includes:
s401, vectorizing the yield data of each hotel to obtain a plurality of characteristic vectors belonging to the hotel. In this embodiment, each of the pieces of yield data includes a feature and a feature value corresponding to the feature. For example, if the production data includes hotel star level data, each piece of hotel star level data includes a star level feature and a feature value corresponding to the feature.
The initial deep learning network model has a plurality of parameters. Each feature matches one of the above parameters. And each feature matches one of the feature vectors. The vectorization operation is performed on the yield data, that is, the vectorization operation is performed on the features in the yield data to obtain feature vectors corresponding to the features. Wherein, the vectorization operation can be implemented by using an embedding technology.
S402, respectively performing characteristic low-order crossing and characteristic high-order crossing on all the characteristic vectors to respectively obtain a first predicted value and a second predicted value. Specifically, as shown in fig. 3, step S402 includes:
s4021, combining any two eigenvectors from all the eigenvectors to form a plurality of eigenvector sets, where each eigenvector set includes two eigenvectors.
S4022, performing inner product operation on the eigenvectors in each eigenvector set respectively to obtain a first predicted value. Specifically, in this step, the inner product operation is performed on the feature vectors in each feature vector set according to the low-order cross function.
The above-mentioned low-order cross-function is:
Figure BDA0002813965160000071
wherein the content of the first and second substances,
Figure BDA0002813965160000072
y1the first predicted value is expressed. w is a0A first parameter representing the initial deep learning network model. w is aiThe ith parameter representing the initial deep learning network model. m represents the total number of features per one of the above yield data. x is the number ofiAnd (4) representing the characteristic value corresponding to the ith characteristic in each piece of the yield data.<vi,vj>Representing a feature vector viAnd vjThe inner product of (d). v. ofiAnd vjRepresenting two eigenvectors in a set of eigenvectors.
viAnd representing a feature vector corresponding to the ith feature in each piece of the yield data. v. ofjAnd representing a feature vector corresponding to the jth feature in each piece of the yield data. x is the number ofjA characteristic value, v, representing the value of the characteristic corresponding to the jth characteristic of each of the aforementioned production datai,fRepresenting a feature vector viThe f-th element of (1), vj,fRepresenting a feature vector vjThe f-th element in (1), k denotes a feature vector viTotal number of elements in (1).
And S4023, obtaining a second predicted value by using all the feature vectors as input to the DNN model. For the calculation process of the DNN model, the prior art is used, and details are not repeated.
And S403, carrying out weighted summation on the first predicted value and the second predicted value to obtain an initial predicted result, and carrying out normalization operation on the initial predicted result by using a preset activation function to obtain a predicted probability value. The weight in the weighted summation can be set according to needs, and the application is not limited. The preset activation function can adopt a softmax activation function.
And S404, taking the prediction probability value and a preset real result value as the input of the quantile loss function to obtain a target network model. Specifically, the loss can be calculated by using the quantile loss function, so that the loss is minimized; at this time, the values corresponding to the parameters in the initial deep learning network model can be obtained, and the characteristic parameters are set to the values, so that the target network model is obtained.
In this embodiment, the target network model includes a quantile, and the quantile is one of the preset quantiles. Quantiles in the target network models corresponding to different hotels are different. Therefore, reasonable quantiles can be set according to data such as attribute characteristics of different hotels, the prediction accuracy of the quantiles is improved, and therefore the prediction accuracy of each hotel is improved.
And S50, respectively predicting the yields of a plurality of hotels in a future preset time period based on the target network model, and outputting the yield values of the hotels in the preset time period. For example, the preset time period may be 6 days, that is, the yield values of each hotel in the future 6 days are predicted respectively. Therefore, the parallel prediction capability of multiple hotels is realized, the time for predicting the yield of the multiple hotels can be saved, and the efficiency of the prediction system for predicting the yield under the multiple hotel scene is improved.
In another embodiment of the present application, based on the above embodiment, as shown in fig. 4, step S404 includes:
s4041, for each hotel, respectively taking a plurality of preset quantiles as the input of the quantile loss function, and training to generate a plurality of different second network models belonging to the same hotel. Each of the second network models corresponds to one of the predetermined quantiles.
S4042, based on the yield data of the past N days as a test set, respectively testing the plurality of different second network models, and using the second network model with the smallest difference between the predicted value and the actual yield value as a target network model corresponding to the hotel. The production data for the past N days includes the true production value.
Therefore, for each hotel, a plurality of standby second network models can be obtained by utilizing a plurality of preset quantiles, the situation that a model convergence partial surface is caused by only selecting one preset quantile generation model is prevented, the obvious influence of abnormal data values on the model can be avoided, the fault tolerance rate of the model is improved, and the generated model is more accurate. And the accuracy of the obtained model parameters can be ensured by using the closest current yield data for testing, and the final prediction result can be more accurate.
In another embodiment of the present application, on the basis of the above embodiment, as shown in fig. 5, in step S40, step S402 further includes: and generating a characteristic ID corresponding to each characteristic according to the characteristics contained in the yield data of each hotel.
Step S403 further includes: and generating a feature dictionary file according to the features and the feature IDs.
Step S404 further includes: and storing the feature dictionary file into the target network model.
Therefore, the problem that in the prior art, a dictionary file is generated in advance according to all the characteristics in a developing process, and the development cost is increased due to the fact that the dictionary file cannot be updated in time under the conditions of newly added characteristics and the like in the subsequent process can be solved, and the characteristic mapping relation does not need to be developed independently, so that the engineering development cost is reduced.
In another embodiment of the present application, on the basis of the above embodiment, as shown in fig. 6, the hotel yield prediction method may further include the steps of:
and S60, updating the yield data of the plurality of hotels based on the hotel yield data of the current time period.
And S70, training the initial deep learning network model again based on the updated yield data to obtain an updated target network model and an updated quantile.
Therefore, the most appropriate quantiles can be found again and updated according to the latest prediction effect, the yield database is enriched, the distribution of data is better described, the timeliness of historical yield data is guaranteed, and the accuracy of next yield prediction is improved.
All the above embodiments of the present application can be freely combined, and the technical solutions obtained by combining them are also within the scope of the present application.
As shown in fig. 7, the embodiment of the present invention further discloses a hotel yield prediction system 7, which includes:
a yield data obtaining module 71, configured to obtain yield data of multiple hotels, where the yield data includes historical yield data, hotel attribute feature data, and current booking progress data;
a loss function construction module 72 for constructing a quantile loss function based on the quantile regression model;
a model construction module 73, configured to construct an initial deep learning network model;
a model training module 74, training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels, and respectively obtaining a target network model corresponding to each hotel;
the yield prediction module 75 predicts the yields of the plurality of hotels in a preset time period in the future based on the target network model, and outputs the yield values of the hotels in the preset time period.
It is understood that the hotel yield prediction system of the present invention further comprises other existing functional modules that support the operation of the hotel yield prediction system. The hotel yield prediction system shown in fig. 7 is only an example, and should not bring any limitations to the function and scope of use of the embodiments of the present invention.
The hotel yield prediction system in this embodiment is used for implementing the method for predicting hotel yield, so for the specific implementation steps of the hotel yield prediction system, reference may be made to the description of the method for predicting hotel yield, and details are not repeated here.
The embodiment of the invention also discloses hotel yield prediction equipment which comprises a processor and a memory, wherein the memory stores the executable instruction of the processor; the processor is configured to perform the steps in the hotel yield prediction method described above via execution of executable instructions. Fig. 8 is a schematic structural diagram of the hotel yield prediction device disclosed by the invention. An electronic device 600 according to this embodiment of the invention is described below with reference to fig. 8. The electronic device 600 shown in fig. 8 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.
As shown in fig. 8, the electronic device 600 is embodied in the form of a general purpose computing device. The components of the electronic device 600 may include, but are not limited to: at least one processing unit 610, at least one memory unit 620, a bus 630 connecting the different platform components (including the memory unit 620 and the processing unit 610), a display unit 640, etc.
Where the memory unit stores program code that may be executed by processing unit 610 to cause processing unit 610 to perform steps according to various exemplary embodiments of the present invention as described in the hotel yield prediction methods section above in this specification. For example, processing unit 610 may perform the steps as shown in fig. 1.
The storage unit 620 may include readable media in the form of volatile memory units, such as a random access memory unit (RAM)6201 and/or a cache memory unit 6202, and may further include a read-only memory unit (ROM) 6203.
The memory unit 620 may also include a program/utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.
Bus 630 may be one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.
The electronic device 600 may also communicate with one or more external devices 700 (e.g., keyboard, pointing device, bluetooth device, etc.), with one or more devices that enable a user to interact with the electronic device 600, and/or with any devices (e.g., router, modem, etc.) that enable the electronic device 600 to communicate with one or more other computing devices. Such communication may occur via an input/output (I/O) interface 650. Also, the electronic device 600 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the Internet) via the network adapter 660. The network adapter 660 may communicate with other modules of the electronic device 600 via the bus 630. It should be appreciated that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms, to name a few.
The invention also discloses a computer readable storage medium for storing a program, which when executed implements the steps of the hotel yield prediction method. In some possible embodiments, the various aspects of the present invention may also be implemented in the form of a program product comprising program code for causing a terminal device to perform the steps according to various exemplary embodiments of the present invention described in the above-mentioned hotel yield prediction method of this specification when the program product is run on the terminal device.
As shown above, when the program of the computer-readable storage medium of this embodiment is executed, the influence of multiple exogenous variables on the final yield can be comprehensively considered, so that the prediction accuracy of the hotel yield is improved; the method has the parallel prediction capability for multiple hotels, can generate models by adopting different quantiles according to different hotels, and ensures the accuracy of predicting the yield of each hotel in parallel prediction on the premise of realizing simultaneous prediction of multiple hotels.
Fig. 9 is a schematic structural diagram of a computer-readable storage medium of the present invention. Referring to fig. 9, a program product 800 for implementing the above method according to an embodiment of the present invention is described, which may employ a portable compact disc read only memory (CD-ROM) and include program code, and may be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited in this regard and, in the present document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
A computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable storage medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).
According to the hotel yield prediction method, the hotel yield prediction system, the hotel yield prediction equipment and the hotel yield prediction storage medium, the influence of a plurality of exogenous variables on the final yield can be comprehensively considered, and the hotel yield prediction accuracy is improved; the method does not depend on historical data for prediction, and can better cope with cold start scenes;
on the other hand, the method and the device have the parallel prediction capability for multiple hotels, different quantiles can be used for generating models according to different hotels, and the accuracy of yield prediction for each hotel in parallel prediction is guaranteed on the premise that simultaneous prediction of the multiple hotels is achieved.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.

Claims (13)

1. A hotel yield prediction method is characterized by comprising the following steps:
obtaining yield data of a plurality of hotels;
constructing a quantile loss function based on the quantile regression model;
constructing an initial deep learning network model;
training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to respectively obtain a target network model corresponding to each hotel;
and respectively predicting the yield of the plurality of hotels in a preset time period in the future based on the target network model, and outputting the yield value of each hotel in the preset time period.
2. The hotel yield prediction method of claim 1, wherein the step of training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to obtain target network models corresponding to the respective hotels comprises:
vectorizing the yield data of each hotel to obtain a plurality of characteristic vectors belonging to the hotel;
respectively performing characteristic low-order crossing and characteristic high-order crossing on the characteristic vector to respectively obtain a first predicted value and a second predicted value;
carrying out weighted summation on the first predicted value and the second predicted value to obtain an initial predicted result, and carrying out normalization operation on the initial predicted result by using a preset activation function to obtain a predicted probability value;
and taking the prediction probability value and a preset real result value as the input of the quantile loss function to obtain a target network model.
3. The hotel yield prediction method of claim 2, wherein the initial deep learning network model comprises a DNN model, and the step of performing feature low-order intersection and feature high-order intersection on the feature vector to obtain the first predicted value and the second predicted value respectively comprises:
optionally combining two feature vectors in all the feature vectors to form a plurality of feature vector sets, wherein each feature vector set comprises two feature vectors;
performing inner product operation on all the characteristic vector sets respectively to obtain a first predicted value;
and taking all the characteristic vectors as the input of the DNN model to obtain a second predicted value.
4. The hotel yield prediction method of claim 1, wherein the target network model comprises quantiles, and the quantiles in the target network models corresponding to different hotels are different.
5. The hotel yield prediction method of claim 1, wherein the step of training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to obtain target network models corresponding to the respective hotels comprises:
for each hotel, respectively taking a plurality of preset quantiles as the input of the quantile loss function, training to generate a plurality of different second network models belonging to the same hotel, wherein each second network model corresponds to one preset quantile;
based on the yield data of the past N days as a test set, respectively testing the plurality of different second network models, and taking the second network model with the smallest difference between the obtained predicted value and the real yield value as a target network model corresponding to the hotel; the production data for the past N days includes the true production value.
6. The hotel yield prediction method of claim 1, wherein the step of training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to obtain target network models corresponding to the respective hotels comprises:
generating a feature ID corresponding to each feature according to the features contained in the yield data of each hotel;
generating a feature dictionary file according to the features and the feature IDs;
storing the feature dictionary file into the target network model.
7. The hotel yield prediction method of claim 3, wherein each of the yield data comprises a characteristic and a characteristic value corresponding to the characteristic; the initial deep learning network model has a plurality of parameters; each of the features matches one of the parameters; and each said feature matches one said feature vector;
the step of performing inner product operation on all the feature vector sets respectively to obtain a first predicted value includes: performing inner product operation on each characteristic vector set according to a low-order cross function;
the low order cross function is:
Figure FDA0002813965150000021
wherein the content of the first and second substances,
Figure FDA0002813965150000022
y1representing said first predicted value, w0A first parameter, w, representing said initial deep learning network modeliAn ith parameter representing the initial deep learning network model, m representing a total number of features per one of the production data, xiRepresenting a feature value corresponding to an ith feature in each of the production data,<vi,vj>representing a feature vector viAnd vjInner product of viA feature vector, v, representing the corresponding ith feature in each of said yield datajA feature vector, x, representing the corresponding jth feature in each of said yield datajRepresenting the jth feature in each of said yield dataCorresponding characteristic value, vi,fRepresenting a feature vector viThe f-th element of (1), vj,fRepresenting a feature vector vjThe f-th element in (1), k denotes a feature vector viTotal number of elements in (1).
8. The hotel yield prediction method of claim 1, wherein the quantile loss function is:
Figure FDA0002813965150000031
wherein n represents the total amount of the production data, q represents a preset quantile, yp ′A predicted value, y, representing said yield data of the p-th slicepThe actual yield value of the p-th piece of the yield data is represented, and L represents the total loss value.
9. The hotel yield prediction method of claim 4, further comprising the steps of:
updating the production data for the plurality of hotels based on the hotel production data for the current time period;
and training the initial deep learning network model again based on the updated yield data to obtain an updated target network model and the updated quantiles.
10. The hotel yield prediction method of claim 1, wherein the yield data comprises historical yield data, hotel attribute feature data, and current booking progress data; the hotel attribute characteristic data comprises hotel star level, house type level and hotel city data, and the yield data further comprises historical booking progress data, current canceling progress, hotel competition circle characteristic and date characteristic.
11. A hotel yield prediction system for implementing the hotel yield prediction method of claim 1, the system comprising:
the system comprises a yield data acquisition module, a reservation module and a reservation module, wherein the yield data acquisition module is used for acquiring yield data of a plurality of hotels, and the yield data comprises historical yield data, hotel attribute characteristic data and current reservation progress data;
the loss function building module is used for building a quantile loss function based on the quantile regression model;
the model building module is used for building an initial deep learning network model;
the model training module is used for training the initial deep learning network model based on the quantile loss function and the yield data of the plurality of hotels to respectively obtain target network models corresponding to the hotels;
and the yield prediction module is used for predicting the yields of the plurality of hotels in a preset time period in the future based on the target network model and outputting the yield values of the hotels in the preset time period.
12. A hotel yield prediction facility, comprising:
a processor;
a memory having stored therein executable instructions of the processor;
wherein the processor is configured to perform the steps of the hotel yield prediction method of any one of claims 1 to 10 via execution of the executable instructions.
13. A computer readable storage medium storing a program, wherein the program when executed by a processor implements the steps of the hotel yield prediction method of any one of claims 1 to 10.
CN202011394090.9A 2020-12-02 2020-12-02 Hotel yield prediction method, system, equipment and storage medium Pending CN112434872A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011394090.9A CN112434872A (en) 2020-12-02 2020-12-02 Hotel yield prediction method, system, equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011394090.9A CN112434872A (en) 2020-12-02 2020-12-02 Hotel yield prediction method, system, equipment and storage medium

Publications (1)

Publication Number Publication Date
CN112434872A true CN112434872A (en) 2021-03-02

Family

ID=74690810

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011394090.9A Pending CN112434872A (en) 2020-12-02 2020-12-02 Hotel yield prediction method, system, equipment and storage medium

Country Status (1)

Country Link
CN (1) CN112434872A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113361920A (en) * 2021-06-04 2021-09-07 上海华客信息科技有限公司 Hotel service optimization index recommendation method, system, equipment and storage medium
CN113496005A (en) * 2021-05-26 2021-10-12 北京房多多信息技术有限公司 Information management method and device, electronic equipment and storage medium

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113496005A (en) * 2021-05-26 2021-10-12 北京房多多信息技术有限公司 Information management method and device, electronic equipment and storage medium
CN113496005B (en) * 2021-05-26 2022-04-08 北京房多多信息技术有限公司 Information management method and device, electronic equipment and storage medium
CN113361920A (en) * 2021-06-04 2021-09-07 上海华客信息科技有限公司 Hotel service optimization index recommendation method, system, equipment and storage medium

Similar Documents

Publication Publication Date Title
CN110766142A (en) Model generation method and device
CN110852438B (en) Model generation method and device
EP4020326A1 (en) Method and apparatus of training model, device, storage medium, and program product
CN110807515A (en) Model generation method and device
WO2018156942A1 (en) Optimizing neural network architectures
KR20190113928A (en) Device placement optimization through reinforcement learning
CN111340221B (en) Neural network structure sampling method and device
JP7262539B2 (en) Conversation recommendation method, device and equipment
CN112465138A (en) Model distillation method, device, storage medium and equipment
CN112434872A (en) Hotel yield prediction method, system, equipment and storage medium
CN111353601A (en) Method and apparatus for predicting delay of model structure
CN116684330A (en) Traffic prediction method, device, equipment and storage medium based on artificial intelligence
CN114358257A (en) Neural network pruning method and device, readable medium and electronic equipment
CN116911574B (en) Three-level supply chain optimization method and device based on whale algorithm and random forest
CN108475346A (en) Neural random access machine
US11256748B2 (en) Complex modeling computational engine optimized to reduce redundant calculations
CN111861801A (en) Hotel full room prediction method, system, equipment and storage medium
CN116703466A (en) System access quantity prediction method based on improved wolf algorithm and related equipment thereof
CN116701975A (en) Wind resource partitioning method and device considering space-time characteristics and electronic equipment
CN116341634A (en) Training method and device for neural structure search model and electronic equipment
CN113642654B (en) Image feature fusion method and device, electronic equipment and storage medium
CN113094602B (en) Hotel recommendation method, system, equipment and medium
CN114897183A (en) Problem data processing method, and deep learning model training method and device
CN112580885A (en) Method, device and equipment for predicting accessory qualification rate and storage medium
CN113128677A (en) Model generation method and device

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