JP6607899B2 - Room charge setting device, guest room charge setting method, and program - Google Patents

Description

  The present invention relates to a guest room charge setting device, a guest room charge setting method, and a program.

  In accommodation facilities, in order to increase profits, systems that support setting of appropriate accommodation charges are used. (For example, Patent Document 1 and Patent Document 2)

JP, 2006-027461, A JP 2016-167305 A

  Patent Documents 1 and 2 describe that an accommodation fee is set based on a booking curve in which a tendency of past accommodation reservations is graphed. However, Patent Documents 1 and 2 disclose a method for setting an appropriate accommodation fee according to various conditions such as a ratio of supply and demand for accommodation facilities, competitiveness for competing accommodation facilities, and seasonal factors. Not.

  Accordingly, an object of the present invention is to automatically set an appropriate accommodation fee corresponding to various conditions.

The accommodation fee setting device according to the present invention is an accommodation fee setting device for setting an appropriate accommodation fee in an accommodation facility,
A base price setting unit for setting the base price of the accommodation fee based on the ratio of the supply amount and demand of the accommodation facility on the first date in the future;
A room charge setting unit that sets a room charge on the first date in the future by verifying the base price set in the machine learning model.

An accommodation fee setting method according to the present invention is an accommodation fee setting method for setting an appropriate accommodation fee in an accommodation facility,
Setting a base price for accommodation based on a ratio of supply and demand for accommodation on a first date in the future;
And setting a room charge on the first date in the future by verifying the base price set in the machine learning model.

The program according to the present invention includes a computer for setting an appropriate room charge at an accommodation facility,
A base price setting unit for setting the base price of the accommodation fee based on the ratio of the supply amount and demand of the accommodation facility on the first date in the future;
By verifying the base price set in the machine learning model, the machine learning model functions as a room charge setting unit that sets the room charge on the first date in the future.

  According to the present invention, it is possible to automatically set appropriate accommodation charges corresponding to various conditions.

The block diagram which shows the structure of the room charge setting apparatus 10 by embodiment of this invention. The figure which shows the example of a booking curve by embodiment of this invention. The flowchart of the training of the machine learning model by embodiment of this invention. The flowchart of the calculation method of the base price by embodiment of this invention. The graph which illustrates the time series change of a demand index and a supply index by embodiment of this invention. The flowchart of the accommodation fee setting process by a machine learning model by embodiment of this invention. The figure which illustrates the process of the setting and verification of the accommodation fee by a machine learning model by embodiment of this invention.

Next, embodiments for carrying out the present invention will be described in detail with reference to the drawings.
FIG. 1 is a block diagram showing a configuration of a guest room charge setting device 10 according to an embodiment of the present invention. As shown in FIG. 1, the guest room charge setting device 10 includes a control device 11 and an external storage device 12. The control device 11 includes, as hardware, a CPU, a memory such as a ROM and a RAM, an input interface, an output interface, a communication interface, a bus connecting these, and the like. The control device 11 implements various functions when the CPU executes a program stored in a ROM or the like. Functional modules realized by the control device 11 include a booking curve extraction unit 111, a machine learning model training unit 112, an accommodation charge setting unit 113, an external factor specifying unit 114, a base price setting unit 115, and an index calculation unit 116. It is. The external storage device 12 is a hard disk drive or the like, and includes a booking curve creation data storage unit 121 and a base price setting information storage unit 122. The guest room charge setting device 10 does not need to be configured by a single computer, and may be configured by a plurality of computers and external storage devices distributed on a communication network. In the present embodiment, the room charge setting device 10 is realized by a cloud service such as SaaS (Software as a Service). In this case, the booking curve creation data storage unit 121 is managed on the cloud server, and each function module (booking curve extraction unit 111, machine learning model training unit 112, accommodation charge setting unit 113, external factor specifying unit 114, base price) The setting unit 115 and the index calculation unit 116) are executed on the cloud server.

  In the booking curve creation data storage unit 121, information on time-series changes in accommodation charges set for a past date (past third date) in a plurality of accommodation facilities (of the accommodation charges on the past date, Information on changes in a given period up to a past date), information on time-series changes in the number of room reservations in a given period up to a past date, and external factors in a given period up to a past date And information about is held. Information on time-series changes in the number of guest room reservations on the past date in a predetermined period until the past date is information that forms a booking curve.

  The base price setting information storage unit 122 stores information generated based on big data collected for setting the base price. Big data includes facility data and macro data. In the facility data, in addition to information on accommodation prices and room inventory, review information on the OTA (Online Travel Agent) site, photos of accommodations posted on the OTA site, facilities and room facilities Information etc. are included. Macro data includes event information (information such as festivals, sports competitions, and other events), seasonal factors (information such as overseas holiday seasons, cherry blossom season), weather, economic statistics (number of visitors to Japan, Exchange rate, etc.), construction data of accommodation facilities, data on private accommodation, behavior data collected from mobile terminals, etc. The base price setting information storage unit 122 stores data (for example, data in a relational database format) obtained by processing various types of big data (NonSQL data) as described above.

  The room charge setting device 10 provides a booking curve used for setting the room charge of accommodation facilities such as hotels and inns. A booking curve is a curve that represents a tendency of an increase in the number of reservations (a decrease in the number of guest rooms) on a specific date. Specifically, for example, as shown in FIG. 2, the horizontal axis of the orthogonal coordinates is a graph created with the number of days until a specific date (current day), and the vertical axis is the number of reservations (room inventory decrease). In general, the slope of the curve becomes steep when the accommodation fee is lower than the appropriate fee, and tends to be gentle when the accommodation fee is high. In addition, various external factors that affect the demand for accommodations (evaluation scores by customers, weather conditions, economic conditions, presence of events, accommodation charges for competing accommodations, reservation status for competing accommodations (rooms The number of stocks) also affects the slope of the booking curve.

  In this embodiment, first, a machine learning model is trained so that an ideal booking curve in the past can be reproduced. Next, a future lodging fee is set by a machine learning model trained using past data, and as a result, the actually obtained booking curve and the past ideal booking curve are compared and verified.

(Training machine learning model using past data)
First, the machine learning model training using the past data by the room charge setting device 10 will be described with reference to the flowchart of FIG.

  First, the guest room charge setting device 10 extracts an ideal booking curve BC (n) (step S101). Specifically, the booking curves of a plurality of accommodation facilities stored in the booking curve creation data storage unit 121 (information on time-series changes in the number of room reservations in the past date in a predetermined period until reaching the past date) From the inside, a booking curve that satisfies a predetermined standard is extracted.

  An ideal booking curve that satisfies a predetermined standard is a booking curve that maximizes the revenue of the accommodation facility. At the same time, it is necessary to take into account the unique characteristics of the accommodation facility, the characteristics of the hotel type, regional characteristics, and the like. For example, in order to maintain the brand value of luxury hotels, even if revenue can be maximized, there is a case where the price of accommodation is not reduced. In addition, accommodation facilities in Okinawa tend to decrease in room inventory at an early stage, so the inclination of the booking curve is very steep. In addition, it is desirable to distribute reservations from the start of sales to the same day in order to hedge the risk that the actual supply and demand situation will deviate significantly. The room charge setting device 10 extracts a case in which a booking curve that satisfies a predetermined criterion set in advance or a booking curve that is closest to the booking curve that satisfies the predetermined criterion is formed. The extraction conditions may differ depending on the accommodation facility (depending on the area, hotel type, etc.). Further, the user may designate a specific case as an ideal booking curve BC (n).

  Next, the guest room charge setting device 10 extracts a non-ideal booking curve BC ′ (n) for comparison with the ideal booking curve BC (n) (step S102). Specifically, an accommodation facility that forms an ideal booking curve BC (n) (here, accommodation facility A), location area, facility type (hotel, guest house, inn, etc.), and room type (Japanese-style room, Western styles, singles, twins, etc.), and other accommodations with the same price range, etc., that do not satisfy the ideal booking curve criteria are extracted. For example, the guest room price setting device 10 may extract the farthest from the ideal booking curve reference. In addition, the extraction conditions may differ depending on the accommodation facility (depending on the region, the hotel type, etc.). Further, the user may designate a specific case as a non-ideal booking curve BC ′ (n).

  Next, the guest room charge setting device 10 extracts an external factor that affects the shape of the booking curve (step S103). Specifically, for example, the correlation between the following external factors (a) to (e) and the inclination of the booking curve and the change in inclination is analyzed, and a certain correlation (for example, | 0.7 | or more) is obtained. Extract recognized external factors. Correlation analysis can be done, for example, for a booking curve that is related to the occurrence of an external factor (for example, a booking curve for a reservation when a specific event is held) and a booking curve that is not related to that external factor. This can be done by comparing the shapes.

(A) Absolute value and fluctuation rate of evaluation score by user calculated by artificial intelligence (b) Accommodation price of competing accommodation facilities, absolute value of room inventory quantity, relative level, fluctuation range, fluctuation speed, etc.
(C) The accommodation charges and the absolute values, relative levels, fluctuation ranges, fluctuation speeds, etc. of private accommodation facilities in the same location area.
(D) Presence / absence of events in the location (e) Season and weather conditions

  Next, the guest room charge setting device 10 labels the training data related to the external factor extracted in step S103 (step S104). The training data is extracted from data stored in the booking curve creation data storage unit 121 (time-series changes in the number of room reservations on a specific day). Specifically, for example, if there is a correlation between the cherry blossom blooming time (external factor Z (n)) and the shape of the booking curve, the data on the time-series changes in the number of room reservations on the day of the cherry blossom blooming date will be labeled It is a target of attachment.

  Next, the guest room price setting device 10 performs training that causes the machine learning model to determine how to change the room charge with respect to external factors, using the training data labeled in step S104 (step S105). . Specifically, for example, the machine learning model is caused to determine whether to increase, maintain, or decrease the accommodation fee on the day when the external factor Z (n) occurs. If the result of the judgment is the same as the ideal booking curve BC (n), the answer is correct. Otherwise (if the choice is the same as the non-ideal booking curve BC '(n)), the answer is incorrect. To do.

  Next, the guest room price setting apparatus 10 evaluates and verifies the machine learning model trained in step S105 using the test data for verification (step S106). The test data is extracted from data stored in the booking curve creation data storage unit 121 (time series change in the number of room reservations on a specific day).

(Base price setting)
Next, a calculation method of a base price applied as an initial value when setting an accommodation fee using the machine learning model trained in steps S101 to S106 (hereinafter referred to as a machine learning model after training) is shown in FIG. This will be described with reference to the flowchart of FIG.

  First, the guest room price setting device 10 calculates various indices using data stored in the base price setting information storage unit 122 (step S111). The index can be calculated using various artificial intelligences and algorithms. Here, as an index, a demand index (indicating demand for daily accommodation facilities in a predetermined period), a supply index (indicating daily supply amount of accommodation facilities in a predetermined period), a competitiveness index (each accommodation facility) Indicate the unique competitiveness).

  The demand index is calculated based on event information (event type, host area, number of spectators, etc.), seasonal factors, weather, economic statistics, behavior data, etc., and the number of guest rooms per day is calculated using artificial intelligence and algorithms.

  The supply index includes data on existing accommodations (price and number of rooms in stock on OTA), construction data on new accommodations (opening date, price, number of rooms, etc.), private property data (location, price, The number of guest rooms supplied per day is estimated based on data such as the number of guest rooms.

  The competitiveness index is determined by comprehensively judging a score based on the review information on the OTA site, a photograph of the accommodation facility posted on the OTA site, and a score based on the facility and the equipment of the guest room. If the review information includes a score (evaluation score, etc.), the score may be used. Further, in the case of only evaluation using sentences or the like, for example, the score may be calculated by performing positive / negative analysis by natural language processing. Moreover, about a photograph, you may make it analyze by image recognition AI and give a score.

  Next, the guest room charge setting device 10 calculates the base price of the guest room charge using the calculated index. First, the room charge setting device 10 has a past date (second date) in which a ratio (SDR) between a demand index and a supply index is closest to a date (first date, hereinafter referred to as the day) for setting a room charge. Is identified (step S112). FIG. 5 is a graph exemplifying a time series change (a daily change) of the demand index and the supply index.

  When the date (D1) that the SDR approximates is specified, the room price setting device 10 determines its own facility to the average of the values obtained by weighting the room unit price of each competing accommodation facility with the facility competitiveness index of each facility on that date. By multiplying the competitiveness index, the base price of the day is calculated (step S113). For example, the room unit price on the date D1 of the competing facility A is a, the unit price on the date D1 of the competing facility B is b, and the unit price on the date D1 of the competing facility C is c, and the facilities A, B, C, and their competition If the power indices are 0.9, 0.7, 1.3, and 1.1, respectively, the base price can be calculated as follows.

Base price = (a × 0.9 + b × 0.7 + c × 1.3) /3×1.1

(Prediction and verification of the future using machine learning models)
Next, using the machine learning model trained in steps S101 to S106 (hereinafter referred to as a machine learning model after training) and the base price calculated in steps S111 to S113, a future room charge is set. The verification process will be described with reference to the flowchart of FIG.

  First, the guest room charge setting device 10 sets a room charge on a specific date in the future (first date) in the machine learning model after training (step S201). Specifically, based on the current data (calculated base price, current number of guest rooms, presence / absence of external factors affecting booking curve), the machine learning model sets the room charge on that date. In particular, when an external factor having a high correlation with the shape of the booking curve occurs, the machine learning model is caused to determine whether to increase, maintain, or decrease the accommodation fee.

  For example, in the case of the cherry blossom blooming time (external factor Z (n)), it is determined to increase the accommodation fee for that period (for example, April 1 to April 10). Further, the range to be increased may be determined based on the correlation coefficient analyzed in step S103. In addition, when it is known that the booking data before and after the same period is also affected, the accommodation fee may be increased for the period before and after.

  Next, after setting the room charge in step S201, the room charge setting device 10 verifies the charge setting based on the machine learning model based on the actual change in the number of guest room inventory thereafter (step S202). Specifically, if the real-time booking curve BC (r) actually obtained for the date on which the accommodation fee is set approximates the ideal booking curve BC (n), the determination is correct. On the other hand, if the slope of BC (r) becomes steeper than that of BC (n), it is determined that the rate of increase in the accommodation fee is insufficient. Further, when the flattening is performed up to a region where the inclination of BC (r) is lower than BC (n), it is determined that the amount of increase is excessive. Even if it is determined that the increase amount is excessive, the accommodation fee may be maintained at least, and may not be decreased.

  The room charge setting device 10 repeats steps S201 to S202 to improve the accuracy of the machine learning model so that BC (r) is as close as possible to BC (n).

  FIG. 7 is a diagram exemplifying a process of setting and verifying a guest room price based on a machine learning model. In the example of FIG. 7, the first price (base price) is set 90 days before the day (the date) on which the accommodation fee is set, and reservation acceptance is started. In the figure, the point indicated by C is the point where the booking fee for the actual booking curve BC (r) deviated from the ideal booking curve BC (n) by more than a certain amount, so the room rate was changed by the machine learning model It is.

  As described above, according to the present embodiment, the base price set by using the machine learning model is set based on the ratio of the supply index and the demand index of the accommodation facility on a certain date in the future. By verifying the above, since an appropriate room charge for a certain day in the future is set, an appropriate room charge corresponding to various conditions can be automatically set.

  Also, identify a certain past day whose ratio of supply index and demand index of the accommodation facility approximates a certain future date, and set the base price based on the accommodation fee of the competing accommodation facility on the past certain day As a result, the base price can be set by a simple and versatile method.

  In addition, since the base price of the accommodation fee is set in consideration of the competitiveness index specific to each accommodation facility, a more accurate and effective base price can be set.

  In addition, since the room charge setting device 10 first trains the machine learning model so that the ideal booking curve in the past can be reproduced, and sets the future room charge by the trained machine learning model, various conditions It is possible to automatically set an appropriate accommodation fee corresponding to the.

  Further, the room charge setting device 10 performs verification by comparing the booking curve actually obtained by the accommodation charge set by the trained machine learning model with the ideal booking curve in the past. The accuracy of the learning model can be further increased.

  In addition, the room charge setting device 10 identifies external factors that influence the time-series changes in the number of guest room reservations, causes the machine learning model to set a room charge that takes into account the external factors, and creates an ideal booking curve. Since the machine learning model is trained so as to obtain the same result, it is possible to automatically set an appropriate room charge considering the occurrence of various external factors.

  The present invention is not limited to the above-described embodiment, and can be implemented in various other forms without departing from the gist of the present invention. For this reason, the said embodiment is only a mere illustration in all points, and is not interpreted limitedly. For example, the above-described processing steps can be arbitrarily changed in order as long as the processing contents do not contradict each other, or can be executed in parallel.

DESCRIPTION OF SYMBOLS 10 ... Accommodation charge setting apparatus 11 ... Control apparatus 12 ... External storage device 111 ... Booking curve extraction part 112 ... Machine learning model training part 113 ... Accommodation charge setting part 114 ... External factor specification part 115 ... Base price setting part 116 ... Index Calculation unit 121 ... Booking curve creation data storage unit 122 ... Base price setting information storage unit

Claims (7)

An accommodation fee setting device for setting an appropriate accommodation fee in an accommodation facility,
An index calculation unit for estimating a supply index indicating a daily supply amount of accommodation facilities in a predetermined period and a demand index indicating a demand for daily accommodation facilities;
A base price setting unit that sets a base price of an accommodation fee based on a ratio between the supply index and the demand index on a first date in the future;
A room charge setting unit configured to set a room charge on the first date in the future by causing the machine learning model to verify the set base price,
The base price setting unit
Identifying a second date in the past having a ratio of the supply index to the demand index on the first date in the future, and a ratio of the closest supply index to the demand index, and competing accommodation on the second date in the past An accommodation fee setting device that sets the base price based on an accommodation fee of a facility . The index calculation unit
Estimate a competitiveness index that indicates the unique competitiveness of accommodation facilities,
The accommodation price setting device according to claim 1 , wherein the base price setting unit sets a base price of an accommodation price based on the competitiveness index. Information on accommodation charges set for a past third date in a plurality of facilities, and the number of room reservations on the past third date in a predetermined period until the past third date Booking curve creation data storage unit that holds information on series change,
A booking curve extraction unit that extracts an ideal booking curve based on information on time-series changes in the number of room reservations on the third date in the past;
A machine learning model training unit that trains the machine learning model to set a room rate for the past third date in the machine learning model and obtain the same result as the ideal booking curve;
The accommodation charge setting unit
The room charge setting device according to claim 1 or 2 , wherein the machine charge model after training is configured to set a room charge on the first date in the future. The machine learning model training unit
Further training of the machine learning model based on the comparison result of the ideal booking curve and the time series change of the number of room reservations on the first date in the future in the period up to the first date in the future The accommodation charge setting device according to claim 3 . An external factor specifying unit for specifying an external factor that affects a time-series change in the number of room reservations on the third date in the past,
The machine learning model training unit
Causing the machine learning model to set a room charge for the past third date in consideration of the external factor, and training the machine learning model to obtain the same result as the ideal booking curve. Item 3. The accommodation charge setting device according to item 3 or 4 . An accommodation fee setting method for setting an appropriate accommodation fee in an accommodation facility by a computer ,
The computer calculates a supply index indicating a daily supply amount of accommodation facilities in a predetermined period, and a demand index indicating a demand for daily accommodation facilities;
The computer sets a base price for accommodation based on a ratio of the supply index to the demand index at a first date in the future;
The computer, the machine learning model, by verifying the base price has been set, only including the step, the to set the rates in the first date of the future,
In the step of setting the base price,
Identifying a second date in the past having a ratio of the supply index to the demand index on the first date in the future, and a ratio of the closest supply index to the demand index, and competing accommodation on the second date in the past A room charge setting method for setting the base price based on a room charge of a facility . A computer to set the proper room rate at the accommodation,
An index calculation unit that estimates a supply index indicating the supply amount of accommodation facilities for each day in a predetermined period and a demand index indicating demand for accommodation facilities for each day;
A base price setting unit that sets a base price of an accommodation fee based on a ratio between the supply index and the demand index on a first date in the future;
By causing the machine learning model to verify the set base price, the machine learning model functions as a room charge setting unit that sets the room charge on the first date in the future ,
The base price setting unit
Identifying a second date in the past having a ratio of the supply index to the demand index on the first date in the future, and a ratio of the closest supply index to the demand index, and competing accommodation on the second date in the past A program for setting the base price on the basis of an accommodation charge of a facility .