Revenue forecasting device, computer system, and revenue forecasting method

Abstract

To provide an income prediction device that improves accuracy of predicting fare income in a future prediction period.SOLUTION: An income prediction device holds actual information indicating actual passenger demand values for each passenger attribute, fare information indicating a fare, model information indicating a model indicating an amount of variation in passenger demand during a prediction period and a passenger attribute to which the model is applied, acquires a parameter value in the prediction period that is input into the model, calculates, on the basis of the model into which the parameter value is input and the actual passenger demand value of the passenger attribute indicated by the model information, a first prediction result indicating the amount of variation in passenger demand during the prediction period for the passenger attribute, and calculates, on the basis of the first prediction result and the fare information, a prediction fare income value during the prediction period.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a revenue prediction device, a computer system, and a revenue prediction method.

Background Art

[0002] As the background art of the present invention, there is Japanese Patent Application Laid-Open No. 2020-098388 (Patent Document 1). This publication states that "The demand prediction device extracts characteristic words indicating the attributes of each product from each document in which the attributes of existing products whose sales have started or new products whose sales have not started are described, based on preset conditions. The demand prediction device generates clustering information indicating a combination of degrees of having characteristic words for each product from the appearance frequencies of the characteristic words included in each document. The demand prediction device sets the generated clustering information as an explanatory variable and learns a prediction model for predicting the demand for new products using learning data in which the actual sales results of existing products are set as the objective variable." (See the abstract).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The technique described in Patent Document 1 performs demand prediction for new products using the past sales results of existing products. However, in predicting the fare revenue of trains, buses, etc., parameters indicating the effects of fare changes and changes in lifestyle in the future prediction period cannot be considered from the past sales results, and there is a risk that the prediction accuracy of the fare revenue in the future prediction period will decrease. Therefore, one aspect of the present invention is to improve the prediction accuracy of the fare revenue in the future prediction period.

Means for Solving the Problems

[0005] To solve the above problems, one aspect of the present invention adopts the following configuration. The revenue forecasting device includes a processor and a memory, the memory holding actual information showing actual passenger demand values ​​for each passenger attribute, fare information showing fares, a model showing the amount of change in passenger demand during the forecast period, and the passenger attributes to which the model is applied, and model information showing models, the processor acquires parameter values ​​for the forecast period that are input to the model, calculates a first forecast result showing the amount of change in passenger demand for the passenger attribute during the forecast period based on the model into which the parameter values ​​have been input and the actual passenger demand values ​​for the passenger attribute shown in the model information, and calculates a forecast fare revenue value for the forecast period based on the first forecast result and the fare information. [Effects of the Invention]

[0006] According to one aspect of the present invention, the accuracy of forecasting fare revenue over a future forecast period can be improved.

[0007] Other issues, configurations, and effects not mentioned above will be clarified by the following description of the embodiments. [Brief explanation of the drawing]

[0008] [Figure 1] This is a block diagram showing an example configuration of the fare revenue forecasting device in Example 1. [Figure 2] This figure shows an example of the data structure of passenger demand performance information in Example 1. [Figure 3] This figure shows an example of the data structure of passenger attribute information in Example 1. [Figure 4] This figure shows an example of the data structure of fare information in Example 1. [Figure 5] This is a flowchart showing an example of the passenger demand forecasting process in Example 1. [Figure 6A] This is an explanatory diagram illustrating an example of the analysis process of passenger demand data to extract trend-indicating features in Example 1. [Figure 6B]This is an explanatory diagram illustrating an example of the analysis process of passenger demand data for extracting features that indicate periodicity in Example 1. [Figure 7A] This figure shows an example of the data structure of the passenger demand forecast results in Example 1. [Figure 7B] This figure shows an example of the data structure of the passenger demand forecast results for each passenger attribute in Example 1. [Figure 8] This flowchart shows an example of the passenger demand fluctuation simulation process in Example 1. [Figure 9] This figure shows an example of the data structure of the simulation model information in Example 1. [Figure 10] This figure shows an example of the data configuration for the simulation conditions in Example 1. [Figure 11A] This is an explanatory diagram showing an example of a commute frequency model in Example 1. [Figure 11B] This is an explanatory diagram showing an example of a behavioral change prediction model in Example 1. [Figure 12] This figure shows an example of the data structure of the simulation result information in Example 1. [Figure 13] This is a flowchart showing an example of the prediction model adjustment process in Example 1. [Figure 14] This figure shows an example of the data structure of the model selection result information in Example 1. [Figure 15] This flowchart shows an example of the fare revenue forecasting process in Example 1. [Figure 16] This figure shows an example of the data structure of the fare revenue forecast result information in Example 1. [Figure 17] This figure shows an example of the screen configuration of the scenario setting screen in Example 1. [Figure 18] This figure shows an example of the screen configuration of the prediction screen in Example 1. [Figure 19] This figure shows an example of the screen configuration of the analysis screen in Example 1. [Figure 20] This is an explanatory diagram illustrating the solution concept of the fare revenue forecasting device in Example 1. [Figure 21A] It is an explanatory diagram showing an example of a use case of the freight revenue prediction device in the first embodiment. [Figure 21B] It is an explanatory diagram showing another example of a use case of the freight revenue prediction device in the first embodiment.

Mode for Carrying Out the Invention

[0009] Hereinafter, embodiments of the present invention will be described in detail based on the drawings. In this embodiment, the same components are generally denoted by the same reference numerals, and repeated explanations will be omitted. It should be noted that this embodiment is merely an example for realizing the present invention and does not limit the technical scope of the present invention.

Examples

[0010] FIG. 1 is a block diagram showing a configuration example of the freight revenue prediction device. The freight revenue prediction device 1 is configured by, for example, a computer having a CPU (Central Processing Unit) 11, an input device 12, an output device 13, a communication device 14, a memory 15, and an auxiliary storage device 16.

[0011] The CPU 11 includes a processor and executes a program stored in the memory 15. The memory 15 includes a ROM (Read Only Memory), which is a non-volatile storage element, and a RAM (Random Access Memory), which is a volatile storage element. The ROM stores invariant programs (such as BIOS (Basic Input / Output System)). The RAM is a high-speed and volatile storage element such as a DRAM (Dynamic Random Access Memory), and temporarily stores the program executed by the CPU 11 and the data used during the execution of the program.

[0012] The auxiliary storage device 16 is a high-capacity, non-volatile storage device such as a magnetic storage device (HDD (Hard Disk Drive)) or flash memory (SSD (Solid State Drive)), and stores the program executed by the CPU 11 and the data used when the program is executed. In other words, the program is read from the auxiliary storage device 16, loaded into memory 15, and executed by the CPU 11.

[0013] The input device 12 is a device that receives input from the operator, such as a keyboard or mouse. The output device 13 is a device that outputs the program execution results in a format that the operator can see, such as a display device or printer.

[0014] The communication device 14 is a network interface device that controls communication with other devices according to a predetermined protocol. The communication device 14 may also include a serial interface such as USB (Universal Serial Bus).

[0015] Some or all of the program executed by the CPU 11 may be provided to the fare revenue forecasting device 1 via a network from a removable media (such as a CD-ROM or flash memory) or an external computer equipped with a non-temporary storage device, and stored in a non-volatile auxiliary storage device 16, which is also a non-temporary storage device. For this reason, the fare revenue forecasting device 1 may have an interface for reading data from the removable media.

[0016] The fare revenue forecasting device 1 is a computer system that operates on a single physical computer or on multiple logically or physically configured computers, and may operate in separate threads on the same computer, or on a virtual computer built on multiple physical computer resources.

[0017] Memory 15 includes, for example, programs: a passenger demand forecasting unit 21, a passenger demand fluctuation simulation unit 22, a forecast model adjustment unit 23, and a fare revenue forecasting unit 24. The passenger demand forecasting unit 21 uses predetermined algorithms such as machine learning and time series forecasting to forecast passenger demand in the future forecast period based on actual passenger demand. The passenger demand fluctuation simulation unit 22 uses a simulation model to simulate the amount of fluctuation in passenger demand in the future forecast period.

[0018] The prediction model adjustment unit 23 determines the distribution ratio (which is also the weight corresponding to each of these prediction results) for the prediction results from the passenger demand forecasting unit 21 and the prediction results from the passenger demand fluctuation simulation unit 22. The fare revenue forecasting unit 24 forecasts fare revenue for the future forecast period based on the prediction results from the passenger demand forecasting unit 21, the prediction results from the passenger demand fluctuation simulation unit 22, and the distribution ratio.

[0019] For example, the CPU 11 implements the passenger demand forecasting function by operating according to the passenger demand forecasting unit 21, which is a program loaded into memory 15, and implements the passenger demand fluctuation simulation function by operating according to the passenger demand fluctuation simulation unit 22, which is a program loaded into memory 15. The relationship between other programs included in memory 15 and the functions implemented by the CPU 11 is similar.

[0020] Furthermore, some or all of the functions implemented by the program contained in memory 15 may be implemented by hardware such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field-Programmable Gate Array).

[0021] The auxiliary storage device 16 stores, for example, passenger demand performance information 31, passenger attribute information 32, fare information 33, passenger demand forecast result information 34, simulation model information 35, simulation conditions 36, simulation result information 37, fare revenue forecast result information 38, and model selection result information 39.

[0022] Passenger demand performance information 31 shows past passenger demand performance. Passenger attribute information 32 shows passenger attributes. Fare information 33 shows past fare performance and future plans. Passenger demand forecast result information 34 shows the forecast results by the passenger demand forecasting unit 21. Simulation model information 35 includes information about the simulation model used for forecasting by the passenger demand fluctuation simulation unit 22.

[0023] Simulation conditions 36 indicate the conditions for the simulation when the passenger demand fluctuation simulation unit 22 uses the simulation model. Simulation result information 37 shows the prediction results by the passenger demand fluctuation simulation unit 22. Fare revenue prediction result information 38 shows the fare revenue prediction results by the fare revenue prediction unit 24. Model selection result information 39 shows the adjustment results (distribution ratio described above) by the prediction model adjustment unit 23.

[0024] Furthermore, some or all of the information stored in the auxiliary storage device 16 may be stored in the memory 15, or in a database connected to the device.

[0025] In this embodiment, the information used by the fare revenue forecasting device 1 is not dependent on a specific data structure and may be represented in any data structure. For example, a data structure appropriately selected from a table, list, database, or queue can store the information.

[0026] The fare revenue forecasting device 1 is connected to an external system 2 and an external server 3 via a network 4 such as the Internet. The external system 2 and the external server 3 transmit information to the fare revenue forecasting device 1 and perform processing based on the information received from the fare revenue forecasting device 1. For example, the fare revenue forecasting device 1 may be integrated with at least one of the external system 2 and the external server 3.

[0027] Figure 2 shows an example of the data structure of passenger demand performance information 31. Passenger demand performance information 31 shows, for example, the actual number of passengers from each departure point to each destination for each time period on each date. Passenger demand performance information 31 is obtained in advance, for example, based on OD (Origin Destination) data acquired by an automatic ticket gate system, which is an example of an external system 2.

[0028] Figure 3 shows an example of the data structure of passenger attribute information 32. Passenger attribute information 32 shows the ratio of passengers of each travel type for each combination of time of day, departure point, and destination. Although a region is defined in passenger attribute information 32, the region is uniquely determined by the combination of departure point and destination. The travel type is information indicating the purpose of the passenger's ride, and includes, for example, commuting, going to school, and travel. Time of day, departure point, destination, region, and travel type are examples of passenger attributes. Passenger attribute information 32 is obtained in advance, for example, according to input from the user of the fare revenue forecasting device 1.

[0029] Figure 4 shows an example of the data structure of fare information 33. Fare information 33 includes fare tables for each period. In this embodiment, a higher number assigned to a fare table indicates a more recent period. Each fare table stores fares for each combination of time period, departure point, and destination. Note that fare information 33 is obtained in advance based on data held by, for example, a railway company's fare management system, which is an example of an external system 2.

[0030] Figure 5 is a flowchart of an example of passenger demand forecasting processing. The passenger demand forecasting unit 21 reads passenger demand actual information 31 (S501). The passenger demand forecasting unit 21 analyzes the passenger demand actual information 31 and extracts feature quantities that indicate the characteristics of passenger demand changes (S502). The passenger demand forecasting unit 21 uses a predetermined algorithm such as machine learning or time series forecasting method (for example, ARIMA (Auto Regressive Integrated Moving Average)) to extract feature quantities that indicate the periodicity of passenger demand changes and feature quantities that indicate the trend of passenger demand changes (for example, the amount of change between consecutive periods (for example, the average difference between the current period and the previous period)).

[0031] Figure 6A is an explanatory diagram illustrating an example of the analysis process of passenger demand performance information 31 for extracting trend-indicating features. The points in Figure 6A represent the total number of passengers over a predetermined period (e.g., one day) as shown in the passenger demand performance information 31, and the graph in Figure 6A is a regression line connecting each point over a predetermined period (e.g., one year). In other words, the graph in Figure 6A shows the time-series change in the actual number of passengers.

[0032] The passenger demand forecasting unit 21 estimates the annual growth rate of passenger demand (number of passengers) as a trend-indicating feature, based on changes in the number of passengers year by year (an example of a predetermined period).

[0033] Figure 6B is an explanatory diagram illustrating an example of the analysis process of passenger demand performance information 31 for extracting features that indicate periodicity. The points in Figure 6B represent the number of passengers over a predetermined period (e.g., one day) as shown in the passenger demand performance information 31, with the trend element calculated in Figure 6A removed (normalized). The graph in Figure 6B is a regression curve fitted so that the same time-series waveform is repeated over a predetermined period (e.g., one week). In other words, the graph in Figure 6B shows the time-series change in the actual number of passengers.

[0034] The passenger demand forecasting unit 21 estimates the day-of-the-week characteristics of passenger demand (number of passengers) (e.g., the rate of increase or decrease for each day of the week) as a feature that indicates periodicity, based on the number of passengers on each day of the week. The passenger demand forecasting unit 21 may also estimate the monthly characteristics of passenger demand (number of passengers) (e.g., the rate of increase or decrease for each month) as a feature that indicates periodicity, based on the number of passengers on each month. Furthermore, the passenger demand forecasting unit 21 may also estimate the seasonal characteristics of passenger demand (number of passengers) (e.g., the rate of increase or decrease for each season) as a feature that indicates periodicity, based on the number of passengers on each season (e.g., spring from March to May, summer from June to August, autumn from September to November, and winter from December to February).

[0035] Furthermore, since both the trend-indicating features and the periodicity-indicating features are characteristics of a specific period (timing), there are corresponding trend-indicating features and periodicity-indicating features for each combination of date and time.

[0036] Returning to the explanation of Figure 5, the passenger demand forecasting unit 21 generates a passenger demand forecast result (described later using Figure 7A) for a future period (for example, specified by the user of the fare revenue forecasting device 1) based on the actual passenger demand information 31 and the features extracted in step S502 (S503).

[0037] The passenger demand forecasting unit 21 generates passenger demand forecast results for each passenger attribute (described later using Figure 7B) based on the demand forecast results generated in step S503 and the passenger attribute information 32, and saves the generated passenger demand forecast results for each passenger attribute and the passenger demand forecast results generated in step S503 to the passenger demand forecasting result information 34 (S504), and terminates the passenger demand forecasting process.

[0038] Figure 7A shows an example of the data structure of the passenger demand forecast results generated in step S503. Figure 7B shows an example of the data structure of the passenger demand forecast results for each passenger attribute generated in step S504. Figures 7A and 7B show an example where passenger demand for 2023 is forecasted based on the actual number of passengers in 2022.

[0039] First, let's explain how the passenger demand forecast result 341 in Figure 7A is generated. The passenger demand forecasting unit 21 retrieves each record of the passenger demand actual information 31. The passenger demand forecasting unit 21 performs the following processing on the top record of the passenger demand actual information 31 in Figure 2 (the record with date "2022 / 1 / 1", time zone "5:00", departure point "A", destination "B", and number of passengers "100").

[0040] The passenger demand forecasting unit 21 stores the date "2023 / 1 / 1", which corresponds to the date of the record "2022 / 1 / 1" in the future (one year later), the time of the record "5:00", the departure point "A", and the destination "B" in the passenger demand forecasting result 431. The passenger demand forecasting unit 21 also stores the number of passengers "100" as the actual value for the previous year in the passenger demand forecasting result 431.

[0041] Furthermore, the passenger demand forecasting unit 21 stores the increase / decrease values ​​from the previous year's actual values, which are caused by the trend-indicating features extracted in step S502, and the increase / decrease values ​​from the previous year's actual values, which are caused by the periodicity-indicating features extracted in step S502, as "trend" and "periodicity" in the figure, respectively, in the passenger demand forecasting result 431.

[0042] Furthermore, the passenger demand forecasting unit 21 calculates a "predicted value (total)" by adding the increase / decrease value due to a trend-indicating feature and the increase / decrease value due to a periodicity-indicating feature to the previous year's actual value, and stores this value in the passenger demand forecast result 431. Through the above process, a record of the future passenger demand forecast result 341 corresponding to the top record of the passenger demand actual information 31 in Figure 2 is generated. The passenger demand forecasting unit 21 performs the above process for all records in the passenger demand actual information 31.

[0043] Next, we will explain how to generate the passenger demand forecast results 342 for each passenger attribute shown in Figure 7B. The passenger demand forecasting unit 21 refers to the passenger attribute information 32 and obtains the ratio of travel types (an example of passenger attributes) for each combination of time of day, departure point, destination, and region.

[0044] The passenger demand forecasting unit 21 calculates the previous year's actual value and the forecast value for each travel type in a given combination by multiplying the ratio of travel types for which the combination of time zone, departure point, and destination indicated in the record matches the record's actual value and forecast value (total) for that record.

[0045] The passenger demand forecasting unit 21 stores the combination of the time period, the departure point, and the destination, the region corresponding to the departure point and the destination, the travel type, the actual value for the previous year, and the demand forecast value in the passenger demand forecasting results 342 for each passenger attribute. Furthermore, the passenger demand forecasting unit 21 stores the ratio of the actual value for the previous year and the demand forecast value as the demand forecast value (compared to the previous year) in the passenger demand forecasting results 342 for each passenger attribute.

[0046] In the example described above, the passenger demand forecasting unit 21 used the ratio of travel types indicated by the passenger attribute information 32 to calculate the demand forecast value from the actual value of the previous year. However, it may also predict the ratio of travel types from past data and use the predicted ratio of travel types. Specifically, for example, the passenger demand forecasting unit 21 predicts the ratio of travel types by determining that passengers who used commuter passes for outbound travel in the morning and evening are commuters, passengers who used student commuter passes for outbound travel in the morning and evening are students, passengers whose travel was combined with Shinkansen travel are tourists, and the portion of travel beyond the regular travel zone for commuter pass users is classified as other travel types. The types of tickets, such as commuter passes, are assumed to be obtained in advance from OD (Origin Destination) data acquired by an automatic ticket gate system, which is an example of an external system 2.

[0047] Figure 8 is a flowchart showing an example of passenger demand fluctuation simulation processing. The passenger demand fluctuation simulation unit 22 registers a simulation model (hereinafter, the simulation model will also be simply called the model) in the simulation model information 35 according to input from the user of the fare revenue forecasting device 1 (S801).

[0048] The passenger demand forecasting unit 21 receives input of simulation conditions for one or more scenarios in accordance with the input from the user of the fare revenue forecasting device 1, and registers them in the simulation conditions 36 (S802). A scenario is defined by a usage model, the current and future parameter values ​​of the usage model, and the distribution ratio described later. Since the simulation conditions 36 do not depend on the distribution ratio, they are defined for each combination of the usage model and the current and future parameter values ​​of the usage model.

[0049] Figure 9 shows an example of the data structure of the simulation model information 35. The simulation model information 35 includes, for example, a model ID that identifies the model, a model name, a prediction type, parameters, and target attributes.

[0050] The prediction type indicates the item predicted by the model. The parameters indicate the parameters of the model. The target attributes indicate the attributes of the passengers that are the target of the model's prediction. The information for each record included in the simulation model information 35 is registered in step S801 according to the input from the user of the fare revenue prediction device 1.

[0051] Figure 10 shows an example of the data structure for simulation condition 36. Simulation condition 36 is registered for each scenario, and Figure 10 shows the simulation condition 36 corresponding to "Scenario 0". Simulation condition 36 includes, for example, the model ID, whether the model is used or not, the model parameters, the current parameter values ​​of the model, and the future parameter values ​​of the model.

[0052] The "Model Usage" field holds information indicating whether the corresponding model is used in the given scenario. The model parameters contain the same items as the parameters corresponding to the model ID in the simulation model information 35. The "Current Parameter Value" indicates the current value of the parameter. The "Future Parameter Value" indicates the value of the parameter in the future forecast period. The information for each record included in the simulation conditions 36 for each scenario is registered in step S802 according to the input from the user of the fare revenue forecasting device 1.

[0053] Returning to the explanation of Figure 8, the passenger demand fluctuation simulation unit 22, for each scenario, refers to the passenger attribute information 32 and extracts the passenger demand forecast targets for each model used in that scenario (S803).

[0054] Specifically, for example, the passenger demand fluctuation simulation unit 22 identifies models for which model use is indicated as "○" in the simulation conditions 36 corresponding to each scenario, and identifies the target attributes in the simulation model information 35 for each identified model. Furthermore, the passenger demand fluctuation simulation unit 22 extracts the passenger demand forecast targets for each model by identifying the number of passengers whose attributes (time of day, travel type, and region) indicated in the passenger attribute information 32 correspond to the target attributes from the number of passengers indicated in the passenger demand actual information 31.

[0055] For example, if the model with model ID "A" shown in Figures 9 and 10 is used, the target attribute for this model is "commuters." Therefore, the passenger demand forecast targets corresponding to the model with model ID "A" are extracted by summing the values ​​obtained by multiplying the number of passengers in each record shown in the passenger demand performance information 31 by the commuter ratio shown in the corresponding record in the passenger attribute information 32.

[0056] Next, the passenger demand fluctuation simulation unit 22 predicts the increase or decrease in passenger demand for each scenario based on the passenger demand forecast target corresponding to the model used in that scenario, the model, the current parameter values ​​of each model, and the future parameter values ​​(S804). The passenger demand fluctuation simulation unit 22 saves the simulation results for each scenario to the simulation result information 37 (S805) and terminates the passenger demand fluctuation simulation process.

[0057] Figure 11A is an explanatory diagram showing an example of a commute frequency model. The commute frequency model is the model with model ID "A" in the simulation model information 35 of Figure 9. The commute frequency model includes a remote work coefficient as a parameter, and the target attribute is "commuter" and the prediction type is "increase / decrease in the number of users". The remote work coefficient is defined, for example, by the ratio of remote work days to the total number of working days for commuters.

[0058] The commute frequency model shows that the lower the remote work coefficient, the more commutes are likely to occur. Therefore, in simulation condition 36 in Figure 10, the current remote work coefficient is "0.5" and the future remote work coefficient is "0.4," which predicts an increase in the number of commutes for commuters in the future, and consequently, an increase in the number of users.

[0059] For example, the current and future parameter values ​​of the remote work coefficient input into the commute frequency model may be determined according to the results of a passenger survey. The commute frequency model is an example of a model that predicts the amount of change in the number of users based on parameter values ​​that indicate changes in passengers' lifestyles.

[0060] Figure 11B is an explanatory diagram showing an example of a behavior change prediction model. The behavior change prediction model is the model with model ID "D" in the simulation model information 35 of Figure 9. The commute frequency model is a model in which "fare difference" is included as a parameter, the target attribute is "region Z", and the prediction type is "change in usage time". The fare difference for each time period in region Z is defined, for example, by subtracting the fare for the time period adjacent to the time period from each fare corresponding to that time period in region Z (the departure and destination corresponding to region Z are identified from the passenger attribute information 32) in the fare table shown in fare information 33.

[0061] The behavioral change prediction model indicates that the larger the fare difference, the higher the probability that the corresponding passenger will change their travel time. Note that passengers who change their travel time are assumed to move to a time slot adjacent to the current time slot, where the fare in the future forecast period is lower. The behavioral change prediction model is an example of a model that predicts changes in travel time based on parameter values ​​indicating fare changes.

[0062] Although not shown in the diagram, the trip frequency model and the price sensitivity model will also be explained. The trip frequency model is the model with model ID "C" in the simulation model information 35 of Figure 9. The price sensitivity model includes "fare" as a parameter, and the target attribute is "traveler" with a prediction type of "increase / decrease in the number of users." The travel coefficient is defined, for example, by the proportion of travel days within a given period for travelers. The trip frequency model shows that the higher the travel coefficient, the more trips are taken, and consequently, the more users there are.

[0063] For example, the current and future parameter values ​​of the travel coefficient input into the travel frequency model may be determined according to the results of a passenger survey. The travel frequency model is an example of a model that predicts the amount of change in the number of users based on parameter values ​​that indicate changes in passengers' lifestyles.

[0064] The price sensitivity model is the model with model ID "C" in the simulation model information 35 of Figure 9. The price sensitivity model includes "fare" as a parameter, and the prediction type for the target attribute "other than commuting to school" is "fare". The price sensitivity model shows that as fares increase, users other than commuting to school will cancel their travel, and consequently, the number of users will decrease. The price sensitivity model is an example of a model that predicts the amount of change in the number of users based on parameter values ​​that indicate changes in fares.

[0065] Figure 12 shows an example of the data structure of the simulation result information 37. The simulation result information 37 is registered for each scenario, and Figure 12 shows the simulation result information 37 corresponding to "Scenario 0". Since the simulation result information 37 does not depend on the distribution ratio described later, it is defined for each combination of the usage model and the current and future parameter values ​​of that usage model.

[0066] The simulation result information 37 shows, for example, the year-on-year change in the forecasted demand and the corresponding model for each combination of time of day, origin, destination, region, and travel type. The corresponding model is the model applied to the combination, i.e., the model used is marked "〇" and corresponds to the attributes shown by the combination. The year-on-year change in the forecasted demand shows the ratio of increase or decrease in the number of passengers over the future forecast period (in this case, one year) due to the application of the corresponding model. The information for each record included in the simulation result information 37 for each scenario is generated in step S804 and registered in step S805.

[0067] Figure 13 is a flowchart showing an example of the prediction model adjustment process. The prediction model adjustment unit 23 obtains prediction results from machine learning, specifically passenger demand prediction results 342 for each passenger attribute (S1301). The prediction model adjustment unit 23 also obtains prediction results from simulation, i.e., simulation result information 37 corresponding to each scenario (S1302).

[0068] The prediction model adjustment unit 23 extracts records for each scenario in which the time period, departure point, destination, region, and travel type overlap, from the passenger demand forecast results 342 for each passenger attribute and the simulation result information 37 corresponding to each scenario (S1303).

[0069] The prediction model adjustment unit 23, for example, in accordance with the input from the user of the fare revenue prediction device 1, selects for each scenario whether to use machine learning prediction results or simulation prediction results for each model used in that scenario (this is equivalent to deciding to set the distribution ratio of the selected prediction method to 100% and the distribution ratio of the unselected prediction method to 0%), or determines the distribution ratio for each prediction method (S1304). The prediction model adjustment unit 23 stores the determined result in the model selection result information 39 (S1305) and terminates the prediction model adjustment process.

[0070] Figure 14 shows an example of the data structure of the model selection result information 39. The model selection result information 39 is registered for each scenario, and Figure 14 shows the model selection result information 39 corresponding to "Scenario 0".

[0071] The model selection result information 39 shows, for example, the distribution ratio of prediction results by machine learning and the distribution ratio of prediction results by simulation for each combination of time of day, origin, destination, region, and travel type.

[0072] For example, the passenger demand forecast results 342 for each passenger attribute in Figure 7B and the simulation result information 37 for Scenario 0 in Figure 12 both contain records corresponding to the time of day "5:00", departure point "A", destination "B", region "X", and travel type "commuter". In other words, in step S1303, the forecast model adjustment unit 23 extracts the relevant records from the passenger demand forecast results 342 and the simulation result information 37. Furthermore, the model corresponding to the relevant record in the simulation result information 37 is Model A.

[0073] Furthermore, in step S1304, the prediction model adjustment unit 23 determined the prediction result by machine learning as the selection result corresponding to model A of scenario 0. Therefore, in the record corresponding to the time period "5:00", departure point "A", destination "B", region "X", and travel type "commute" in the model selection result information 39 of Figure 14, it stored "100%" as the machine learning value and "0%" as the simulation value.

[0074] Furthermore, for example, the passenger demand forecast results 342 for each passenger attribute in Figure 7B and the simulation result information 37 for Scenario 0 in Figure 12 both include records corresponding to the time zone "7:00", departure point "E", destination "F", region "Z", and travel type "travel". In other words, in step S1303, the forecast model adjustment unit 23 extracts the relevant records from the passenger demand forecast results 342 and the simulation result information 37. The models corresponding to the relevant records in the simulation result information 37 are Model B and Model D.

[0075] Furthermore, in step S1304, the prediction model adjustment unit 23 determined that "50%" was the distribution ratio for the prediction results obtained by machine learning and "50%" was the distribution ratio for the prediction results obtained by simulation, corresponding to models B and D of scenario 0. Therefore, in the record corresponding to the time zone "7:00", departure point "E", destination "F", region "Z", and travel type "travel" in the model selection result information 39 in Figure 14, "50%" was stored as the machine learning value and "50%" as the simulation value.

[0076] Furthermore, in the simulation result information 37 for Scenario 0 in Figure 12, the model for time zone "6:00", departure point "C", destination "D", region "Y", and travel type "commuting to school" is "No model", and the prediction results from the simulation cannot be used. Therefore, in the record corresponding to time zone "6:00", departure point "C", destination "D", region "Y", and travel type "commuting to school" in the model selection result information 39 in Figure 14, "100%" is forcibly stored as the machine learning value and a null value (equivalent to 0%) as the simulation value.

[0077] Figure 15 is a flowchart showing an example of fare revenue forecasting processing. The fare revenue forecasting unit 24 reads the future forecast period (S1501). The fare revenue forecasting unit 24 reads the passenger demand forecast results 342 for each passenger attribute, the simulation result information 37 for each scenario, and the model selection result information 39 for each scenario (S1502). The fare revenue forecasting unit 24 reads the fare information 33 (S1503).

[0078] The fare revenue forecasting unit 24 calculates the fare revenue for each scenario (S1504), stores the information indicating the calculation results in the fare revenue forecasting result information 38 (S1505), and terminates the fare revenue forecasting process.

[0079] Figure 16 shows an example of the data structure of the fare revenue forecast result information 38. The fare revenue forecast result information 38 is generated in step S1504 and saved in step S1505. The fare revenue forecast result information 38 includes information showing the year-on-year change in fares, previous year's actual values, forecast values, and demand forecast values ​​corresponding to each combination of date, time zone, origin, destination, region, and travel type included in the forecast period, for each scenario, and information showing the revenue forecast value for each date included in the forecast period for each scenario.

[0080] In the example in Figure 4, the fare information 33 is assumed to be uniformly defined, but as shown in the example in Figure 16, the fare may differ depending on the scenario (i.e., the fare information 33 may be defined for each scenario).

[0081] The following explains an example of the calculation process for the fare revenue forecast result information 38 for Scenario 0 in Figure 16, using the example of the passenger demand forecast result 342 for each passenger attribute in Figure 7B, the example of the simulation result information 37 for Scenario 0 in Figure 12, and the example of the model selection result information 39 for Scenario 0 in Figure 14.

[0082] The fare revenue forecasting unit 24 stores the date, time zone, origin, destination, region, travel type, and previous year's actual value columns of the passenger demand forecasting results 342 for each passenger attribute shown in Figure 7B in a table that displays the fare revenue forecasting results for Scenario 0. Furthermore, the fare revenue forecasting unit 24 refers to the fare information 33 and stores the fares corresponding to each combination of time zone, origin, and destination in the table.

[0083] Furthermore, according to the model selection result information 39 for Scenario 0 in Figure 14, in Scenario 0, the allocation ratio for machine learning is 100% and the allocation ratio for simulation is 0% for the attributes specified by time zone "5:00", departure point "A", destination "B", region "X", and travel type "commute" (i.e., for these attributes, only the passenger demand forecast results 342 for each passenger attribute are used). Also, since the year-on-year change in the demand forecast value for this attribute in the passenger demand forecast results 342 for each passenger attribute in Figure 7B is +10%, +10% is adopted as the year-on-year change in the demand forecast value corresponding to this attribute in Scenario 0.

[0084] Similarly, according to the model selection result information 39 for Scenario 0 in Figure 14, in Scenario 0, the allocation ratio for machine learning for the attribute specified by time zone "6:00", departure point "C", destination "D", region "Y", and travel type "commuting to school" is 100%, and the allocation ratio for simulation is null (equivalent to 0%). In addition, the year-on-year change in the passenger demand forecast result 342 for each passenger attribute in Figure 7B is +20%, so +20% is adopted as the year-on-year change in the demand forecast value corresponding to that attribute in Scenario 0.

[0085] Furthermore, according to the model selection result information 39 for Scenario 0 in Figure 14, in Scenario 0, the allocation ratio for machine learning and the allocation ratio for simulation are 50% for the attributes specified by time zone "7:00", departure point "E", destination "F", region "Z", and travel type "travel". Also, the year-on-year change in the passenger demand forecast result 342 for each passenger attribute in Figure 7B is -10%, and the year-on-year change in the demand forecast value for the same attribute in the simulation result information 37 for Scenario 0 in Figure 12 is +20%. Therefore, the year-on-year change in the demand forecast value corresponding to the attribute in Scenario 0 is adopted as -10% × 50% + 20% × 50% = 5%.

[0086] The fare revenue forecasting unit 24 calculates the forecast value by multiplying the previous year's actual value by the year-on-year change in the corresponding demand forecast value. For each date in Scenario 0, the fare revenue forecasting unit 24 calculates the fare revenue forecast for each date in Scenario 0 by calculating the sum of the products of the fare and the forecast value.

[0087] By performing the above-described process for Scenario 0 for all other scenarios, the fare revenue for each scenario in step S1504 is calculated.

[0088] Figure 17 shows an example of the screen configuration of the scenario setting screen. The scenario setting screen 1700 is displayed, for example, on the output device 13. The scenario setting screen 1700 includes, for example, a simulation condition setting area 1710, a model registration area 1720, and a prediction model adjustment area 1730.

[0089] The simulation condition setting area 1710 includes, for example, a pull-down menu 1711, buttons 1712 and 1713, and a simulation condition table 1714. The pull-down menu 1711 allows selection of the scenario to be set. Selecting button 1712 adds a new scenario. Selecting button 1713 allows the addition of a fare table corresponding to the currently set scenario. The simulation condition table 1714 is an area for registering the simulation conditions 36 for the currently set scenario.

[0090] For example, if the user specifies the parameter values ​​for each model in a given scenario, the passenger demand fluctuation simulation unit 22 may automatically generate multiple scenarios by increasing or decreasing the values ​​of the specified parameters within a predetermined range and width.

[0091] The model registration area 1720 includes, for example, a pull-down menu 1721, a button 1722, a model display area 1723, and a summary field 1724. The pull-down menu 1721 allows selection of the model to be set. When button 1722 is selected, a new model is added to the simulation model information 35. The model display area 1723 displays a function indicating the model being set. The summary field 1724 allows input of a summary of the model being set. The summary entered in the summary field 1724 may be registered in the simulation model information 35.

[0092] The prediction model adjustment area 1730 includes, for example, a pull-down menu 1731, a checkbox 1732, a pull-down menu 1733, a selection result display area 1734, and a button 1735. The pull-down menu 1731 allows selection of the model for which the allocation ratio in the configured scenario will be determined. When the checkbox 1732 is selected, the allocation ratio of the simulation of that model in the configured scenario is set to 100%. The pull-down menu 1733 allows selection of the allocation ratio of machine learning and simulation of that model for each combination of origin and destination in the configured scenario. When the button 1735 is selected, the allocation ratio determined by the selection of checkbox 1732 or pull-down menu 1733 is stored in the model selection result information 39.

[0093] Figure 18 shows an example of the screen configuration of the prediction screen. The prediction screen 1800 is displayed, for example, on the output device 13. The prediction screen 1800 includes, for example, a graph display area 1810, a period input area 1820, and a prediction result display area 1830.

[0094] The graph display area 1810 displays graphs showing the fare revenue forecast for each date in each scenario indicated by the fare revenue forecast result information 38, and the actual fare revenue based on the actual values ​​for each date.

[0095] The period input area 1820 includes dropdown menus that accept input for, for example, the period for displaying actual fare revenue and the future forecast period. Furthermore, when the "Add Simulation Scenario" button in the period input area 1820 is selected, the user transitions to the scenario settings screen 1700.

[0096] The forecast result display area 1830 displays, for example, a summary of the fare revenue forecast for each scenario shown in the fare revenue forecast result information 38 (for example, the total number of days in the forecast period for each scenario, the average daily revenue during the forecast period, and the total revenue during the forecast period). Also, when the "Compare / Analyze Scenarios" button in the forecast result display area 1830 is selected, the user transitions to the analysis screen described later.

[0097] Figure 19 shows an example of the screen configuration of the analysis screen. The analysis screen 1900 is displayed, for example, on the output device 13. The analysis screen 1900 includes, for example, a pull-down menu 1901, a pull-down menu 1902, a passenger demand forecast display area 1903, a pull-down menu 1904, and a model fluctuation forecast display area 1905.

[0098] The departure point can be selected using pull-down menu 1901. The destination can be selected using pull-down menu 1902. The passenger demand forecast display area 1903 displays records from the passenger demand forecast results 341 that correspond to the departure point selected in pull-down menu 1901 and the destination selected in pull-down menu 1902.

[0099] A scenario can be selected using the pull-down menu 1904. The model fluctuation forecast display area 1905 displays the fluctuation amount of the demand forecast value for each time period of the scenario selected in the pull-down menu 1904, with each model as the factor, as well as the sum of the fluctuation amounts of the demand forecast values ​​for all models (total fluctuation forecast value).

[0100] The amount of change in the demand forecast value for each time period of the selected scenario, which is displayed in the model fluctuation prediction display area 1905 and is caused by each model, is calculated, for example, by multiplying the demand forecast value (compared to the previous year) for each model in the simulation result information 37 of the scenario for that time period by the actual value for the previous year shown in the passenger demand forecast result information 34 for the attribute corresponding to that model.

[0101] The analysis screen 1900 allows users of the fare revenue forecasting device 1 to view predictions made by algorithms such as machine learning, as well as the fluctuations caused by each model (i.e., factors that cause demand to increase or decrease).

[0102] Figure 20 is an explanatory diagram illustrating the solution concept of the fare revenue forecasting device 1. The fare revenue forecasting device 1 receives information from the user, such as passenger data, fare data, and model parameters for each scenario. Passenger demand performance information 31, which shows the number of passengers obtained from OD data, and passenger attribute information 32 are both examples of passenger data. Fare information 33, which shows actual and planned fares, is an example of fare data.

[0103] Furthermore, the fare revenue forecasting device 1 is pre-equipped with machine learning algorithms such as Prophet, ARIMA, and DNN (Deep Neural Network), as well as time series forecasting algorithms. In addition, the fare revenue forecasting device 1 is equipped with simulation models such as the price sensitivity model, commute frequency model, and travel frequency model mentioned above.

[0104] The fare revenue forecasting device 1 performs demand forecasting using machine learning and time-series forecasting based on actual data such as passenger data. Furthermore, the fare revenue forecasting device 1 creates scenarios that assume changes in fares or lifestyles, for example, according to user input, and predicts demand fluctuations in those scenarios. The fare revenue forecasting device 1 performs a fare revenue simulation that combines the demand forecast and the demand fluctuations in those scenarios, and reports the forecast results by displaying the forecast screen 1800 and the analysis screen 1900, etc.

[0105] Figure 21A is an explanatory diagram illustrating an example of a use case for the fare revenue forecasting device 1. In the example shown in Figure 21A, the fare revenue forecasting device 1 is included in the operation planning system 2100. The operation planning system 2100 further includes an operation planning device 2101, which is an example of an external server 3, and a fare management system 2102, which is an example of an external system 2.

[0106] The operation planning device 2101, for example, plans the operation of mobility such as trains and buses that passengers ride (e.g., timetables). The fare management system 2102 generates a fare table. The fare revenue forecasting device 1 uses the operation plan created by the operation planning device 2101 to forecast passenger demand. The fare revenue forecasting device 1 also stores the fare table generated by the fare management system 2102 in the fare information 33.

[0107] The fare revenue forecasting device 1 transmits fare revenue forecasting result information 38 to the operation planning device 2101 and the fare management system 2102. The operation planning device 2101 updates the operation plan based on the received fare revenue forecasting result information 38. Specifically, for example, the operation planning device 2101 increases the number of mobility services for time periods and areas where the predicted number of passengers indicated by the fare revenue forecasting result information 38 is determined to be high based on predetermined conditions, or decreases the number of mobility services for time periods and areas where the predicted number of passengers indicated by the fare revenue forecasting result information 38 is determined to be low based on predetermined conditions (for example, a function showing a relationship where the number of services increases as the predicted number of passengers per time period increases is predetermined, and the number of services corresponding to the predicted number of passengers indicated by the fare revenue forecasting result information 38 is adopted).

[0108] The fare management system 2102 updates the fare table based on the received fare revenue forecast result information 38. Specifically, for example, the fare management system 2102 raises fares for time periods and regions where the fare revenue forecast result information 38 indicates a high predicted number of passengers based on predetermined conditions, or lowers fares for time periods and regions where the fare revenue forecast result information 38 indicates a low predicted number of passengers based on predetermined conditions (for example, a function is predetermined that shows a relationship where fares increase as the predicted number of passengers per time period increases, and fares corresponding to the predicted number of passengers indicated by the fare revenue forecast result information 38 are adopted).

[0109] The fare revenue forecasting device 1 may receive the updated operation plan from the operation planning device 2101 and the updated fare table from the fare management system 2102, and may regenerate the fare revenue forecasting result information 38 based on the updated operation plan and the updated fare table.

[0110] According to the operation planning system 2100 in Figure 21A, the operation plan and fares can be optimized based on the fare revenue forecast by the fare revenue forecasting device 1 (for example, the profitability of the transportation operator that owns the mobility can be improved).

[0111] Figure 21B is an explanatory diagram illustrating an example of a use case for the fare revenue forecasting device 1. In the example in Figure 21B, the fare revenue forecasting device 1 is included in the management support system 2110. The management support system 2110 further includes a management simulator 2111, which is an example of an external server 3, and an asset management system 2112, which is an example of an external system 2.

[0112] The management simulator 2111 performs simulations related to the management of a transportation operator, for example. The management simulator 2111 transmits to the fare revenue forecasting device 1 information obtained from the simulation, such as information on fare increases and decreases, and campaign information (which affects model parameters, for example). The asset management system 2112 manages the assets (including mobility such as buses and trains) owned by the transportation operator. The asset management system 2112 transmits to the fare revenue forecasting device 1 information such as reductions in service area due to accidents or construction (information that may cause unexpected revenue losses).

[0113] The fare revenue forecasting device 1 transmits fare revenue forecasting result information 38 to the management simulator 2111 and the asset management system 2112. The management simulator 2111 makes management decisions based on the received fare revenue forecasting result information 38. Specifically, for example, if the future forecast fare revenue indicated by the fare revenue forecasting result information 38 is higher than a predetermined value, the management simulator 2111 makes management decisions such as starting a new business or increasing personnel costs. If the future forecast fare revenue indicated by the fare revenue forecasting result information 38 is lower than a predetermined value, the management simulator makes management decisions such as stopping an existing business or reducing personnel costs.

[0114] The asset management system 2112 manages the estimates held by the transportation operator based on the received fare revenue forecast result information 38. Specifically, for example, if the projected future fare revenue indicated by the fare revenue forecast result information 38 is higher than a predetermined value, the asset management system 2112 makes a decision to increase the amount of capital investment, and if the projected future fare revenue indicated by the fare revenue forecast result information 38 is lower than a predetermined value, it makes a decision to decrease the amount of capital investment.

[0115] According to the management support system 2110 in Figure 21B, management decisions and asset management can be optimized based on fare revenue forecasts from the fare revenue forecasting device 1 (for example, the profitability of transportation operators that own mobility services can be improved).

[0116] As described above, the fare revenue forecasting device 1 of this embodiment combines demand forecasting using machine learning based on past performance data with demand forecasting using a simulation model. By using machine learning based on past performance data for demand forecasting, the fare revenue forecasting device 1 can reflect trends and periodicities based on past performance in the forecast values. Furthermore, by performing demand forecasting using a simulation model, the fare revenue forecasting device 1 can reflect factors that have not been previously seen in the forecast values. In addition, by performing forecasting model adjustment processing, the fare revenue forecasting device 1 can make fare revenue forecasts under various conditions.

[0117] In this embodiment, the fare revenue forecasting device 1 is described as performing a forecast that combines demand forecasting using machine learning based on past performance data with demand forecasting using a simulation model. However, only one of these demand forecasts may be used.

[0118] It should be noted that the present invention is not limited to the embodiments described above, and various modifications are included. For example, the embodiments described above are described in detail to make the present invention easier to understand, and are not necessarily limited to those having all the configurations described. It is also possible to replace parts of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add configurations from other embodiments to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace parts of the configuration of each embodiment with other configurations.

[0119] Furthermore, each of the above configurations, functions, processing units, and processing means may be implemented in hardware, either partially or entirely, by designing them as integrated circuits, for example. Alternatively, each of the above configurations and functions may be implemented in software by having the processor interpret and execute programs that implement each function. Information such as programs, tables, and files that implement each function can be stored in memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

[0120] Furthermore, the control lines and information lines shown are those deemed necessary for explanatory purposes, and not all control lines and information lines are necessarily shown in the actual product. In reality, it is safe to assume that almost all components are interconnected. [Explanation of Symbols]

[0121] 1 Fare Revenue Forecasting Device, 2 External System, 3 External Server, 11 CPU, 12 Input Device, 13 Output Device, 14 Communication Device, 15 Memory, 16 Auxiliary Storage Device, 21 Passenger Demand Forecasting Unit, 22 Passenger Demand Fluctuation Simulation Unit, 23 Forecast Model Adjustment Unit, 24 Fare Revenue Forecasting Unit, 31 Passenger Demand Actual Information, 32 Passenger Attribute Information, 33 Fare Information, 34 Passenger Demand Forecasting Result Information, 35 Simulation Model Information, 36 Simulation Conditions, 37 Simulation Result Information, 38 Fare Revenue Forecasting Result Information, 39 Model Selection Result Information, 1700 Scenario Setting Screen, 1800 Forecasting Screen, 1900 Analysis Screen, 2100 Operation Planning System, 2101 Operation Planning Device, 2102 Fare Management System, 2110 Management Support System, 2111 Management Simulator, 2112 Asset Management System

Claims

1. It is an income forecasting device, Including the processor and memory, The aforementioned memory is Actual passenger demand data showing passenger demand figures by passenger attribute, Fare information showing the fare, The system maintains a model that shows the fluctuations in passenger demand during the forecast period, and model information that shows the passenger attributes to which the model is applied. The aforementioned processor, The parameter values ​​for the prediction period input to the model are obtained, Based on the model into which the parameter values ​​have been input and the actual passenger demand values ​​for the passenger attributes indicated by the model information, a first forecast result is calculated that shows the amount of change in passenger demand for the passenger attributes during the forecast period. A revenue forecasting device that calculates a predicted fare revenue value for the forecast period based on the first forecast result and the fare information.

2. A revenue forecasting device according to claim 1, A revenue forecasting device in which the parameter values ​​input to the model include at least one of a value indicating the passenger's lifestyle and a value indicating the fare.

3. A revenue forecasting device according to claim 1, The aforementioned performance information shows the actual passenger demand values ​​for each of the aforementioned passenger attributes for a predetermined period. The aforementioned processor, Based on a predetermined machine learning or predetermined time series forecasting algorithm and the actual passenger demand values, a second forecast result showing passenger demand during the forecast period is calculated. A revenue forecasting device that calculates a predicted fare revenue value for the forecast period based on the first prediction result, the second prediction result, and the fare information.

4. The revenue forecasting device according to claim 3, The aforementioned processor, The distribution ratio of the first prediction result and the second prediction result for each passenger attribute is obtained, A revenue forecasting device that calculates a predicted fare revenue value for the forecast period based on the distribution ratio, the first forecast result, the second forecast result, and the fare information.

5. The revenue forecasting device according to claim 4, The processor generates data for displaying a screen that accepts inputs of the model, the parameter values, and the distribution ratio, and is used as an income forecasting device.

6. The revenue forecasting device according to claim 4, The aforementioned processor, It accepts input of parameter values ​​and distribution ratios corresponding to each of multiple scenarios. For each of the aforementioned scenarios, Based on the model into which parameter values ​​corresponding to the scenario have been input, and the actual passenger demand values ​​for the passenger attributes indicated by the model information, the first prediction result is calculated. Based on the distribution ratio corresponding to the scenario, the first prediction result corresponding to the scenario, the second prediction result, and the fare information, the estimated fare revenue for the forecast period of the scenario is calculated. A revenue forecasting device that generates data for displaying the predicted fare revenue values ​​for each of the aforementioned multiple scenarios during the forecast period.

7. The revenue forecasting device according to claim 3, The aforementioned processor, Based on the predetermined machine learning or predetermined time series forecasting algorithm and the actual passenger demand values ​​for each passenger attribute, the trend and periodicity of passenger demand for each passenger attribute are predicted. A revenue forecasting device that generates data for displaying the trends and periodicities of passenger demand for each of the predicted passenger attributes.

8. A computer system, Including an income forecasting device and an external device, The revenue forecasting device is Actual passenger demand data showing passenger demand figures by passenger attribute, Fare information showing the fare, The system maintains a model that shows the fluctuations in passenger demand during the forecast period, and model information that shows the passenger attributes to which the model is applied. The parameter values ​​for the prediction period input to the model are obtained, Based on the model into which the parameter values ​​have been input and the actual passenger demand values ​​for the passenger attributes indicated by the model information, a first forecast result is calculated that shows the amount of change in passenger demand for the passenger attributes during the forecast period. Based on the first prediction result and the fare information, the estimated fare revenue for the prediction period is calculated. A computer system that transmits the calculated fare revenue forecast to the external device.

9. A computer system according to claim 8, The aforementioned external device includes a flight planning device, The aforementioned operation planning device is a computer system that determines the number of mobility services to be operated based on the fare revenue forecast value received from the revenue forecasting device.

10. A computer system according to claim 8, The external device includes an asset management system, The asset management system is a computer system that determines changes to the amount of capital investment made by a transportation operator that owns the mobility used by passengers, based on the fare revenue forecast value received from the revenue forecasting device.

11. A method for predicting revenue using a revenue prediction device, The revenue forecasting device includes a processor and memory, The aforementioned memory is Actual passenger demand data showing passenger demand figures by passenger attribute, Fare information showing the fare, The system maintains a model that shows the fluctuations in passenger demand during the forecast period, and model information that shows the passenger attributes to which the model is applied. The aforementioned revenue forecasting method is: The processor acquires the parameter values ​​for the prediction period that are input to the model, The processor calculates a first forecast result indicating the amount of change in passenger demand for a given passenger attribute during the forecast period, based on the model into which the parameter values ​​have been input and the actual passenger demand values ​​for the passenger attribute indicated by the model information. A revenue forecasting method comprising the processor calculating a forecast value of fare revenue for the forecast period based on the first forecast result and the fare information.

Images