JP7449985B2 - Prediction device, prediction method, and program - Google Patents

Description

The present invention relates to a prediction device, a prediction method, and a program.

2. Description of the Related Art In recent years, information processing devices have been used to provide useful information to users who participate in public competitions such as horse races, bicycle races, and boat races. In this information processing device, for example, before a race, prediction information predicting the finish order of competing objects (for example, racehorses, etc.) is provided to support users who are trying to purchase voting tickets.

Recently, information processing apparatuses that not only predict the finish order (that is, race results) but also predict race development information have been known. For example, Patent Document 1 discloses an information processing device that displays individual player information, race development prediction information, and predicted finish order in association with each other.

JP2017-83976A

In the information processing device described in Patent Document 1 mentioned above, the "lines" set during a bicycle race are displayed as development prediction information. This line is a tactic unique to bicycle racing, and refers to groups of athletes running in a line during a race. The information processing device of Patent Document 1 only displays the predicted configuration of each line, and does not provide any information on how the ranking (positioning) of each line will change during the race. Not displayed.

In other words, the information processing device of Patent Document 1 only displays the line configuration specific to bicycle bicycle races as expected development information, and therefore the information is of no use in other publicly managed races. Moreover, since nothing is displayed regarding the positioning of the lines, etc., it is difficult to say that the information processing device of Patent Document 1 predicts the actual development of the race.
For example, in horse racing, the ranking (positioning) of each racehorse in the first three furlongs (600 meters remaining to the finish line) is often considered important. Therefore, showing how each racehorse's ranking changes before and after that, or from the beginning to the middle, can be said to be the real development of the race.

For this reason, there has been a need for a technology that can predict the actual development of a race in publicly managed races and notify users in an easy-to-understand manner.

The present invention has been made in view of the above-mentioned circumstances, and it is an object of the present invention to provide a prediction device, a prediction method, and a program that can predict the actual race development and notify the user in an easy-to-understand manner.

The prediction device according to the first aspect of the present invention includes:
A prediction device for predicting a race result for a race in which a plurality of competitive bodies move along a route in which a plurality of passing points including a goal are set,
a derivation unit that inputs values for a plurality of parameter items regarding each competition object into a learned prediction model, and derives a score value for each competition object for each passing point;
an estimating unit that estimates the ranking of each competition object for each passing point based on each score value derived by the deriving unit;
an output unit that outputs prediction information including a change in the ranking of each of the sports objects in the form of an image or sound based on the ranking of each passing point estimated by the estimating unit;
It is characterized by having the following.

Further, the prediction device according to the above viewpoint is
further comprising a specifying unit that identifies a passing point at which the ranking of a sport object of interest among the plurality of sport bodies changes based on the ranking of each passing point estimated by the estimating unit;
The output unit may further output explanatory information regarding the attention game object at the passing point specified by the identification unit.

Further, the prediction device according to the above viewpoint is
further comprising a reception unit that receives from a user any parameter item included in the plurality of parameter items and a coefficient value for weighting;
The derivation unit may multiply the value of the parameter item by the coefficient value only for the parameter item accepted by the reception unit, and input the multiplied value to the prediction model to derive the score value.

Further, the prediction device according to the above viewpoint is
Each parameter item is adjusted by adjusting the value of each parameter item input into the prediction model and verifying in advance the process of change in the ranking of each sport according to the derived score value. further comprising a selection unit that selects a unit width of each of the coefficient values,
The accepting unit may accept the coefficient value according to the corresponding unit width based on each unit width selected by the selecting unit and the accepted parameter item.

Further, the prediction device according to the above viewpoint is
a first correct answer rate indicating the correct answer rate for the ranking of each sport body estimated without changing the value of each parameter item for past races; the parameter item accepted by the reception unit; further comprising a calculation unit that calculates a second correct answer rate indicating a correct answer rate for the ranking of each sport body estimated by changing the value of the target parameter item according to the coefficient value,
The output unit may further output comparison information that compares the first correct answer rate and the second correct answer rate calculated by the calculation unit.

The prediction method according to the second aspect of the present invention is
A prediction method executed by a prediction device that predicts a race result for a race in which a plurality of competitive bodies move along a route in which a plurality of passing points including a goal are set,
a derivation step of inputting values for a plurality of parameter items regarding each competition object into a learned prediction model, and deriving a score value for each competition object for each passing point;
an estimating step of estimating the ranking of each competition object for each passing point based on each score value derived in the deriving step;
an output step of outputting prediction information including a change in the ranking of each of the sports objects in the form of an image or sound based on the ranking of each passing point estimated in the estimation step;
It is characterized by having the following.

The program according to the third aspect of the present invention is
A computer that predicts race results for races in which multiple athletes move along routes with multiple passing points including the finish line.
a derivation unit that inputs values for a plurality of parameter items regarding each competition object into a learned prediction model, and derives a score value for each competition object for each passing point;
an estimating unit that estimates the ranking of each competition object for each passing point based on each score value derived by the deriving unit;
an output unit that outputs prediction information including a change in the ranking of each competition object in the form of an image or sound based on the ranking of each passing point estimated by the estimating unit;
It is characterized by functioning as

The above program can be distributed and sold via a computer communication network independently of the computer on which the program is executed.
Furthermore, the information recording medium on which the program is recorded is a non-transitory recording medium, and can be distributed and sold independently of the computer. Here, the non-temporary recording medium refers to a tangible recording medium. Examples of non-temporary recording media include compact disks, flexible disks, hard disks, magneto-optical disks, digital video disks, magnetic tapes, semiconductor memories, and the like. Further, the term "transitory recording medium" refers to the transmission medium (propagation signal) itself. The temporary recording medium is, for example, an electrical signal, an optical signal, an electromagnetic wave, or the like. Note that the temporary storage area is an area for temporarily storing data and programs, and is, for example, volatile memory such as RAM (Random Access Memory).

According to the present invention, it is possible to predict a substantial race development and notify the user in an easy-to-understand manner.

FIG. 1 is a block diagram showing an example of the overall configuration of an information providing system according to the present embodiment. FIG. 1 is a block diagram illustrating an example of a schematic configuration of a typical information processing device in which a prediction device and the like according to the present embodiment are implemented. 1 is a block diagram showing an example of a functional configuration of a prediction device according to a first embodiment; FIG. FIG. 2 is a schematic diagram for explaining a prediction model. It is a schematic diagram showing an example of race information. It is a schematic diagram showing an example of racehorse information. FIG. 3 is a schematic diagram showing an example of user information. 3 is a formula diagram showing an example of a prediction screen according to the first embodiment. FIG. 7 is a formula diagram showing an example of another prediction screen according to the first embodiment. FIG. 7 is a flowchart for explaining prediction processing according to the first embodiment. FIG. 3 is a block diagram illustrating an example of a functional configuration of a prediction device according to a second embodiment. 7 is a formula diagram showing an example of a prediction screen according to Embodiment 2. FIG. 7 is a formula diagram showing an example of another prediction screen according to the second embodiment. FIG. 7 is a flowchart for explaining prediction explanation processing according to the second embodiment.

Embodiments of the present invention will be described below. In the embodiment of the present invention, an information provision system for predicting the race development of horse racing, which is one of the publicly managed sports, will be described. However, it is not limited to such horse racing, and as will be explained later, The present invention can be similarly applied to an information provision system for predicting the development of races in publicly managed competitions.
Note that the following embodiments are for illustrative purposes only, and do not limit the scope of the present invention. Therefore, those skilled in the art can adopt embodiments in which any or all of these elements are replaced with equivalents, and these embodiments are also included within the scope of the present invention.

(overall structure)
An information providing system 100 according to an embodiment of the present invention is, for example, a system that provides a user with prediction information including changes in the ranking of each racehorse in a horse race, and as shown in FIG. 200 and a user terminal 300. The prediction device 200 and the user terminal 300 are communicably connected via the Internet 900.
Although only a few user terminals 300 are shown in FIG. 1, it is assumed that the actual information providing system 100 includes a huge number of user terminals 300 depending on the number of users.

Forecasting device 200 is, for example, a server device, and provides useful information to a user who intends to purchase a voting ticket in response to access from user terminal 300.
For example, before the race starts, the prediction device 200 predicts the ranking of each racehorse for each passing point in the race (for example, the 1st corner, 2nd corner, 3rd corner, 4th corner, and the finish line). . Note that the passing points may be each furlong (signs every 200 meters). The prediction device 200 then provides the user with prediction information including changes in the rankings of each racehorse in the form of images or audio.

The user terminal 300 is, for example, a smartphone, a tablet, a personal computer, etc., and is used by a user who uses the information providing system 100.
Note that, as an example, a case will be described in which a dedicated application (application) for using the information providing system 100 is installed on the user terminal 300, but it is also possible to access the prediction device 200 from a general-purpose browser or the like. It may be possible.
When a user who intends to purchase a voting ticket launches a dedicated application on the user terminal 300 and logs in, the user terminal 300 becomes able to access the prediction device 200, including the race development predicted by the prediction device 200. It becomes possible to receive provision of forecast information.

(Summary configuration of information processing device)
A typical information processing device 400 that implements the prediction device 200 and the like according to this embodiment will be described.

As shown in FIG. 2, the information processing device 400 includes a CPU (Central Processing Unit) 401, a ROM (Read Only Memory) 402, a RAM (Random Access Memory) 403, an NIC (Network Interface Card) 404, and an image processor. It includes a processing section 405, an audio processing section 406, an auxiliary storage section 407, an interface 408, an operation unit 409, a display unit 410, and a speaker 411.

The CPU 401 controls the overall operation of the information processing device 400, and is connected to each component to exchange control signals and data.

The ROM 402 stores an IPL (Initial Program Loader) that is executed immediately after the power is turned on, and when this is executed, the program stored in the auxiliary storage unit 407 is read out to the RAM 403 and the program is executed by the CPU 401. Execution begins.

The RAM 403 is for temporarily storing data and programs, and holds programs and data read from the auxiliary storage unit 407, as well as other data necessary for communication.

The NIC 404 is for connecting the information processing device 400 to a computer communication network such as the Internet, and may be one that conforms to the 10BASE-T/100BASE-T standard used when configuring a LAN (Local Area Network). Consists of an analog modem for connecting to the Internet using a telephone line, an ISDN (Integrated Services Digital Network) modem, an ADSL (Asymmetric Digital Subscriber Line) modem, a cable modem for connecting to the Internet using a cable television line, etc. be done.

The image processing unit 405 processes the image data read from the auxiliary storage unit 407 etc. using the CPU 401 and an image calculation processor (not shown) included in the image processing unit 405, and then processes the image data. It is recorded in a frame memory (not shown). The image information recorded in the frame memory is converted into a video signal at predetermined synchronization timing and output to the display unit 410 via the interface 408 or the like. That is, under the control of the CPU 401, the image processing unit 405 generates an image that is necessary during the progress of the processing performed by the information processing apparatus 400, and causes the display unit 410 to display the image.

The audio processing unit 406 converts the audio data read from the auxiliary storage unit 407 and the like into an audio signal, and outputs the audio signal to the speaker 411 via the interface 408. That is, under the control of the CPU 401, the audio processing unit 406 generates audio to be generated during the progress of processing performed by the information processing device 400, and outputs the audio from the speaker 411.

The auxiliary storage unit 407 is a hard disk or the like, and stores various programs and various data necessary for controlling the operation of the entire information processing device 400. For example, the auxiliary storage unit 407 stores a program for implementing the prediction device 200 according to this embodiment. Under the control of the CPU 401, the auxiliary storage unit 407 reads the stored programs and data as appropriate, and temporarily stores them in the RAM 403 or the like.

An operation unit 409, a display unit 410, and a speaker 411 are connected to the interface 408. Note that the interface 408 may transmit and receive necessary information to and from other connected external devices.

The operation unit 409 receives operation input from a worker or the like who uses the information processing device 400 .

The display unit 410 draws an image according to the image data output by the image processing unit 405 and presents it to a worker using the information processing device 400 or the like.

The speaker 411 outputs audio according to the audio data output by the audio processing unit 406, and presents the audio to a worker using the information processing device 400.

In addition, the information processing device 400 may include a drive unit such as a DVD (Digital Versatile Disc)-ROM instead of the auxiliary storage unit 407. In this case, the information processing device 400 reads programs and data from a DVD-ROM etc. installed in the drive unit and operates in the same manner as described above.

(Embodiment 1)
The functional configuration and the like of the prediction device 200 according to the first embodiment will be described below with reference to FIGS. 3 to 9. This prediction device 200 is realized by the information processing device 400 described above. For example, when the information processing device 400 is powered on, a program for functioning as the prediction device 200 is executed, and the prediction device 200 according to the first embodiment is realized.
Note that the user terminal 300 is similarly realized by the information processing device 400, but a description thereof will be omitted, and the most characteristic prediction device 200 in the first embodiment will be described below.

(Functional configuration of prediction device according to Embodiment 1)
FIG. 3 is a block diagram showing an example of the functional configuration of the prediction device 200 according to the first embodiment. As illustrated, the prediction device 200 includes a transmitting section 210, a receiving section 220, a storage section 230, and a control section 240.

The transmitter 210 transmits necessary information to the user terminal 300.
For example, the transmission unit 210 transmits prediction information (prediction information including changes in the rankings of each racehorse during the race) output by the control unit 240 (more specifically, the output unit 244 described later) to the user terminal 300. .

Note that the above-mentioned NIC 404 and the like can function as such a transmitter 210.

The receiving unit 220 receives various information sent from the user terminal 300.
For example, as will be described later, the receiving unit 220 selects a parameter item whose weighting the user desires to change among a plurality of parameter items input into a prediction model (prediction model 231 described below), and multiplies the value of that parameter item. from the user terminal 300.

Note that the above-mentioned NIC 404 and the like can function as such a receiving unit 220.

The storage unit 230 stores various information necessary for predicting race developments and the like. For example, the storage unit 230 stores a prediction model 231, race information 232, racehorse information 233, and user information 234.

First, the prediction model 231 will be explained. The prediction model 231 is, for example, a trained model that is trained using learning data (combined data of input and results) based on past race results.
For example, as shown in FIG. 4, a prediction model 231 is obtained by inputting learning data DT into a neural network NN and learning it. The learning data DT can be input, for example, by horse weight, previous race time (time in the previous race), previous race 3F (time taken from the remaining 600 meters to the finish line in the previous race), etc. These are the values of various representative parameter items. In addition, the results of the learning data DT include, for example, race time (time in that race), furlong time (time between furlongs in that race), and uphill 4F (time required from the remaining 800 meters to the goal in that race). , uphill 3rd floor (time required from the remaining 600 meters to the finish line in the race), corner passing order (passing order at each corner in the race), etc.

Next, the race information 232 will be explained. The race information 232 is information about a race before the race, and includes, for example, a race ID 232a, a main race track 232b, a distance 232c, and a number of horses 232d, as shown in FIG. Note that the race information 232 may include the weather information of the day (temperature, humidity, wind force, etc.), the condition of the horse track (fair, slightly heavy, heavy, poor, etc.), and the like.

Next, the racehorse information 233 will be explained. The racehorse information 233 is information about racehorses participating in the race, and for example, as shown in FIG. Halon) 233e is included. Note that the racehorse information 233 may include the horse name, the horse number at the time of the race, information about the jockey riding the horse at the time of the race, and the like.

Next, the user information 234 will be explained. The user information 234 is information about the user who uses the user terminal 300, and includes, for example, a user ID 234a, a name 234b, a specified item 234c, and a coefficient value 234d, as shown in FIG. Note that the specified item 234c is information for specifying a parameter item that the user wants to weight among the parameter items (horse weight, previous race time, previous race 3F, etc.) input to the prediction model 231 described above. Further, the coefficient value 234d is a weighting value for the parameter item shown in the above specified item 234c.

In addition to this, the storage unit 230 stores graphic data for generating a screen, voice data for making a virtual character, which will be described later, speak, and the like.

Note that the above-mentioned auxiliary storage unit 407 and the like can function as such a storage unit 230.

Returning to FIG. 3, the control unit 240 controls the entire prediction device 200. This control section 240 includes a reception section 241 , a derivation section 242 , an estimation section 243 , and an output section 244 .

The receiving unit 241 receives the parameter items and coefficient values that the receiving unit 220 receives from the user terminal 300. That is, the receiving unit 241 receives a parameter item whose weighting the user desires to change, among various parameter items input to the prediction model 231, and a coefficient value by which the value of the parameter item is multiplied.
Note that if the received coefficient value is greater than 1, this means that the user attaches importance to that parameter item. On the other hand, if the coefficient value is smaller than 1, it means that the user is neglecting the parameter item.

The derivation unit 242 inputs values for various parameter items (for example, horse weight, previous race time, previous race 3F, etc.) regarding each racehorse participating in the race into the learned prediction model 231, and determines passing points. A score value for each racehorse is derived for each racehorse (for example, the first corner, the second corner, the third corner, the fourth corner, and the finish line). Note that the passing points may be each furlong.
In addition, when the receiving unit 241 accepts parameter items and coefficient values, the deriving unit 242 multiplies the value of the parameter item by the received coefficient value only for the accepted parameter items. , is input to the prediction model 231 to derive a score value. In other words, the score value is derived by the derivation unit 242 with the values of the parameter items that the user attaches importance to (or deemphasizes) being modified as appropriate. Thereby, the user can obtain the score value of each racehorse and the ranking, which will be described later, for each passing point based on the AI (Artificial Intelligence) prediction customized by the user.

The estimating unit 243 estimates the ranking of each racehorse for each passing point based on each score value derived by the deriving unit 242 described above. That is, the estimation unit 243 estimates the ranking of each racehorse at each passing point by sorting the racehorses in descending order of score value.

The output unit 244 outputs prediction information including changes in the rankings of each racehorse in the form of images or sounds based on the rankings for each passing point estimated by the estimating unit 243 described above.
For example, the output unit 244 outputs a predicted screen PG1 as shown in FIG. 8, and provides it to the user terminal 300 via the transmitting unit 210. This prediction screen PG1 includes a finish prediction EA and a development prediction PE. The predicted finishing order EA indicates the predicted finishing order of each racehorse when it reaches the goal. On the other hand, the predicted development PE represents the ranking of each racehorse at each passing point (front, 2nd corner: 2nd corner, 3rd corner: 3rd corner, 4th corner: 4th corner, result: goal) by the length of the bar graph GF. (first come, first served). With such a prediction screen PG1 displayed on the user terminal 300 (display unit), the user can, for example, tap each tab TB in order to see how the ranking of each racehorse changes at each passing point. From this, you can get a concrete idea of the expected race development. Note that the tabs TB may be switched in order at regular intervals according to a script or the like set on the prediction screen PG1.

In addition to this, the output unit 244 may output a prediction screen PG2 as shown in FIG. 9, for example, and provide it to the user terminal 300 via the transmission unit 210. This prediction screen PG2 includes a finish prediction EA and a development prediction PE. Note that the development prediction PE on the prediction screen PG2 is performed by the virtual character VC by vocalization (utterance). For example, the virtual character VC will notify the user of the expected development of the race by reading out text as shown in the speech bubble SB (for example, text indicating the ranking of each racehorse at the passing points).

Note that the CPU 401 and the like described above can function as the control unit 240 having such a configuration.

(Operation of prediction device according to Embodiment 1)
The operation of the prediction device 200 according to the first embodiment having such a configuration will be described below with reference to FIG. 10. FIG. 10 is a flowchart showing the flow of prediction processing executed by the prediction device 200. This prediction process is started, for example, when the AI prediction page is accessed from the user terminal 300 and the racetrack and race number are specified.

First, the prediction device 200 determines whether or not user settings are accepted (step S11). For example, the control unit 240 (reception unit 241) determines whether parameter items and coefficient values have been received from the user terminal 300.

When the prediction device 200 determines that the user settings are not accepted (step S11; No), the prediction device 200 advances the process to step S13, which will be described later.

On the other hand, if it is determined that the user settings are accepted (step S11; Yes), the prediction device 200 changes the value of the parameter item (step S12). For example, the control unit 240 changes the value of the received parameter item by multiplying the value of the parameter item by the received coefficient value. Note that if the received coefficient value is greater than 1, this means that the user attaches importance to that parameter item. Conversely, if the coefficient value is smaller than 1, it means that the user is neglecting that parameter item.

The prediction device 200 derives the score value of each racehorse for each passing point (step S13). That is, the control unit 240 (deriving unit 242) uses the predictive model 231 that has learned the values of various parameter items (for example, horse weight, previous race time, previous race 3F, etc.) regarding each racehorse participating in the race. , and a score value for each racehorse is derived for each passing point (for example, the 1st corner, 2nd corner, 3rd corner, 4th corner, and the finish line). Note that the passing points may be each furlong.
Note that when the value of the received parameter item has been changed in step S12 above, the control unit 240 inputs the changed value to the prediction model 231 only for that parameter item and sets the score value. Derive.

The prediction device 200 estimates the ranking of each racehorse for each passing point (step S14). That is, the control unit 240 (estimation unit 243) estimates the ranking of each racehorse for each passing point based on each score value derived in step S13 above. That is, the control unit 240 estimates the ranking of each racehorse at each passing point by sorting each racehorse in descending order of score value.

The prediction device 200 outputs prediction information including changes in the rankings of each racehorse (step S15). That is, the control unit 240 (output unit 244) outputs prediction information including changes in the ranking of each racehorse in the form of images or sounds based on the ranking for each passing point estimated in step S14 above. .
For example, the control unit 240 outputs the prediction screen PG1 as shown in FIG. 8 described above, and provides it to the user terminal 300 via the transmission unit 210. This prediction screen PG1 includes a development prediction PE. This predicted development PE represents the ranking of each racehorse at each passing point (front, second corner, third corner, fourth corner, result) by the length of the bar graph GF. With such a prediction screen PG1 displayed on the user terminal 300, the user can, for example, tap each tab TB in order to check the prediction based on the transition of the ranking of each racehorse at each passing point. You can learn more about the race development in detail. Note that the tabs TB may be switched in order at regular intervals according to a script or the like set on the prediction screen PG1.
In addition to this, the control unit 240 may output the prediction screen PG2 as shown in FIG. 9 described above and provide it to the user terminal 300 via the transmission unit 210. This prediction screen PG2 also includes the development prediction PE. In this development prediction PE, the virtual character VC notifies the user of the predicted race development by reading out text as shown in the speech bubble SB.

Through such prediction processing, the user is informed of the changes in the ranking of each racehorse for each passing point, and the user can specifically know the predicted development of the race.
Further, the score value is derived with the value of the parameter item that the user attaches importance to (or neglects) being modified as appropriate. Thereby, the user can obtain the score value and ranking of each racehorse for each passing point based on the AI prediction customized by the user.

As a result, it is possible to predict the actual race development and notify the user in an easy-to-understand manner.

(Embodiment 2)
In the above-described first embodiment, a case has been described in which notification is given of changes in the rankings of all racehorses participating in a race. Changes in the ranking of the horse of interest may be reported. Note that the horse of interest is, for example, a racehorse marked as a favorite horse in a race, a rival horse, etc., or a racehorse that has been registered as a favorite by the user.
Embodiment 2, which is characterized by notifying changes in the ranking of the horse of interest, will be described below.

(Functional configuration of prediction device according to Embodiment 2)
FIG. 11 is a block diagram showing an example of the functional configuration of the prediction device 200 according to the second embodiment. As illustrated, the prediction device 200 includes a transmitting section 210, a receiving section 220, a storage section 230, and a control section 250.
Note that the transmitting section 210, receiving section 220, and storage section 230 have the same configuration as the transmitting section 210, receiving section 220, and storage section 230 shown in FIG. 3 described above.

The control section 250 includes a receiving section 241 , a deriving section 242 , an estimating section 243 , a specifying section 251 , and an output section 244 . In other words, the control section 250 has a configuration in which the specifying section 251 is added to the control section 240 shown in FIG. 3 described above.

The specifying unit 251 specifies a passing point at which the ranking of the horse of interest among the plurality of racehorses participating in the race changes, based on the ranking for each passing point estimated by the estimating unit 243. For example, for each passing point, the specifying unit 251 compares the ranking of the horse of interest at that passing point with the ranking of the horse of interest at the previous passing point, and if the difference in ranking is greater than a certain value (for example, the ranking is 2 (If the ranking has increased by two or more positions, or if the ranking has decreased by two or more positions), the passing point is specified.

Then, the output unit 244 outputs explanatory information about the horse of interest at the passing point specified by the identification unit 251 as an image or audio.
For example, the output unit 244 outputs a predicted screen PG3 as shown in FIG. 12 and provides it to the user terminal 300 via the transmitting unit 210. This prediction screen PG3 includes a finish prediction EA and a development prediction PE. In addition, in the development prediction PE on the prediction screen PG3, among the bar graphs GF showing the ranking of each racehorse at the passing point (as an example, corner 3) specified by the specifying unit 251, the horse of interest (for example, No. 5) is selected. Only the bar graph FG of ○○ Rufu) is highlighted.
In addition to this, the output unit 244 may output a prediction screen PG4 as shown in FIG. 13, for example, and provide it to the user terminal 300 via the transmission unit 210. This prediction screen PG4 includes a finish prediction EA and a development prediction PE. In addition, in the development prediction PE on the prediction screen PG4, the virtual character VC reads aloud a text as shown in the speech bubble SB (for example, a text informing the ranking of the horse of interest at the passing point specified by the specifying unit 251). , users will be informed of race developments focusing on horses of interest.

Note that the CPU 401 and the like described above can function as the control unit 250 having such a configuration.

(Operation of prediction device according to Embodiment 2)
The operation of the prediction device 200 according to the first embodiment having such a configuration will be described below with reference to FIG. 14. FIG. 14 is a flowchart showing the flow of the prediction explanation process executed by the prediction device 200. This prediction explanation process is started, for example, when the AI prediction page is accessed from the user terminal 300 and the racetrack and race number are specified.
Note that steps S11 to S14 in the prediction explanation process in FIG. 14 are the same processes as steps S11 to S14 in the prediction process in FIG. 10 described above. Therefore, steps S11 to S14 will be briefly explained.

First, the prediction device 200 determines whether or not user settings are accepted (step S11). When the prediction device 200 determines that the user settings are not accepted (step S11; No), the prediction device 200 advances the process to step S13, which will be described later. On the other hand, if it is determined that the user settings are accepted (step S11; Yes), the prediction device 200 changes the value of the parameter item (step S12).

The prediction device 200 derives the score value of each racehorse for each passing point (step S13). The prediction device 200 estimates the ranking of each racehorse for each passing point (step S14).

The prediction device 200 identifies passing points at which the ranking of the horse of interest changes (step S21). That is, the control unit 250 (specification unit 251) determines the passing point at which the ranking of the horse of interest among the plurality of racehorses participating in the race changes based on the ranking for each passing point estimated in step S14 above. Identify. For example, for each passing point, the control unit 250 compares the ranking of the horse of interest at that passing point with the ranking of the horse of interest at the previous passing point, and if the difference in ranking is greater than a certain value, the control unit 250 specifies the passing point. do.

The prediction device 200 provides explanatory information about the horse of interest at the specified passing point (step S22). That is, the control unit 250 (output unit 244) outputs explanatory information about the horse of interest at the passing point specified in step S21 above in the form of images or audio.
For example, the control unit 250 outputs the prediction screen PG3 as shown in FIG. 12 described above, and provides it to the user terminal 300 via the transmission unit 210. This prediction screen PG3 includes a development prediction PE. In this development prediction PE, only the bar graph FG of the horse of interest is highlighted among the bar graphs GF showing the ranking of each racehorse at the specified passing points.
In addition to this, the control unit 250 may output the prediction screen PG4 as shown in FIG. 13 described above and provide it to the user terminal 300 via the transmission unit 210. This prediction screen PG4 also includes the development prediction PE. In this development prediction PE, the virtual character VC reads out text as shown in the speech bubble SB to inform the user of the development of the race focusing on the horse of interest.

Through such prediction commentary processing, the user is informed of the changes in the ranking of the horse of interest at the passing points, which are the highlights, and the user can specifically learn about the development of the race focusing on the horse of interest.

As a result, it is possible to predict the actual race development and notify the user in an easy-to-understand manner.

(Other embodiments)
In the first and second embodiments described above, the case where the reception unit 241 accepts an arbitrary coefficient value has been described, but after the control units 240 and 250 select an appropriate unit width, the user requests a coefficient value according to the unit width. A coefficient value may also be accepted.
For example, the control units 240 and 250 may further include a selection unit. This selection section adjusts the values of various parameter items input to the prediction model 231, and verifies in advance the process of change in the ranking of each racehorse according to the derived score value. Select the unit width of the coefficient value corresponding to each parameter item.
In the present invention, instead of directly predicting the racehorse's ranking, the racehorse's score value is derived and the racehorse's ranking is estimated in the order of the score value, so the values of the parameter items input to the prediction model 231 There is a range in which the ranking of racehorses hardly changes even if . This width can be the unit width of the coefficient value.
Note that, as an example of a degree in which there is almost no change, it is possible to adopt a condition in which the rankings match 99%.

To explain more specifically, if increasing the coefficient value of a certain parameter item to 1.12 causes almost no change, the selection unit selects 0.12 as the unit width on the plus side. Furthermore, if the coefficient value of the same parameter item is reduced to 0.85, the selection unit selects 0.15 as the unit width on the negative side if the coefficient value does not change substantially.
Then, based on each unit width selected by the selection unit and the accepted parameter item, the reception unit 241 receives a coefficient value according to the corresponding unit width.

Thereby, appropriate coefficient values can be received from the user according to the parameter items.

In the first and second embodiments described above, a case has been described in which when the user changes the value of a parameter item, there is no notification of any difference from the case where the value of the parameter item is not changed. The user may be notified of information comparing the case where the value of the parameter item is not changed and the case where the value of the parameter item is changed.
For example, the control units 240 and 250 may further include a calculation unit. This calculation unit calculates, for past races, a first correct answer rate indicating the correct answer rate for the ranking of each racehorse estimated without changing the value of each parameter item, the parameter item accepted by the receiving unit 241, and , and a second correct answer rate indicating the correct answer rate for the estimated ranking of each racehorse by changing the value of the target parameter item according to the coefficient value.
That is, the calculation unit first estimates the actual ranking at each passing point in the past race and each score value obtained by inputting the value of each parameter item into the prediction model 231 without changing it. The proportion of matches (first correct answer rate) is calculated. Next, the calculation unit changes the value of the corresponding parameter item according to the actual ranking, the parameter item received by the reception unit 241, and the coefficient value, and inputs it into the prediction model 231 to obtain each score value. The rate of agreement (second correct rate) with the estimated rank based on the calculated ranking is calculated.
Then, the output unit 244 outputs comparison information that compares the first correct answer rate and the second correct answer rate calculated by the calculation unit.

Thereby, it is possible to notify the user whether the parameter items and coefficient values changed by the user are appropriate.

In Embodiments 1 and 2 above, a case has been described in which prediction information such as the race development predicted by the user by changing the value of an arbitrary parameter item is provided to the user. The prediction information may be made public to other users.

In the above embodiment, the information providing system 100 for horse racing was described as an example of a publicly managed competition, but the present invention is also applicable to other publicly managed competitions such as boat races and bicycle races.

As described above, according to the present invention, it is possible to provide a prediction device, a prediction method, and a program that can predict a substantial race development and notify the user in an easy-to-understand manner.

100 Information provision system 200 Prediction device 210 Transmission section 220 Receiving section 230 Storage section 231 Prediction model 232 Race information 233 Racehorse information 234 User information 240, 250 Control section 241 Reception section 242 Derivation section 243 Estimation section 244 Output section 251 Specification section 300 User terminal 400 Information processing device 401 CPU
402 ROM
403 RAM
404 NIC
405 Image processing section 406 Audio processing section 407 Auxiliary storage section 408 Interface 409 Operation unit 410 Display unit 411 Speaker 900 Internet

Claims (7)

A prediction device for predicting a race result for a race in which a plurality of competitive bodies move along a route in which a plurality of passing points including a goal are set,
a derivation unit that inputs values for a plurality of parameter items regarding each competition object into a learned prediction model, and derives a score value for each competition object for each passing point;
an estimating unit that estimates the ranking of each competition object for each passing point based on each score value derived by the deriving unit;
an output unit that outputs prediction information including a change in the ranking of each of the sports objects in the form of an image or sound based on the ranking of each passing point estimated by the estimating unit ;
The score value is a value that allows estimating the ranking of each competition object by sorting in descending order,
The estimating unit estimates the ranking of each of the competition objects for each passing point by sorting the score values of each of the competition objects in descending order,
The prediction model is a trained model that has been trained to output the score value for each of the sports objects for each passing point when values for a plurality of parameter items regarding each of the sports objects are input. ,
A prediction device characterized by : further comprising a specifying unit that identifies a passing point at which the ranking of a sport object of interest among the plurality of sport bodies changes based on the ranking of each passing point estimated by the estimating unit;
The output unit emphasizes and outputs the competition object of interest at the passing point specified by the identification unit in the prediction information .
The prediction device according to claim 1, characterized in that: further comprising a reception unit that receives from a user any parameter item included in the plurality of parameter items and a coefficient value for weighting;
The derivation unit multiplies the value of the parameter item by the coefficient value only for the parameter item accepted by the reception unit, and inputs the multiplied value to the prediction model to derive the score value.
The prediction device according to claim 1 or 2, characterized in that: Each parameter item is adjusted by adjusting the value of each parameter item input into the prediction model and verifying in advance the process of change in the ranking of each sport according to the derived score value. further comprising a selection unit that selects a unit width of each of the coefficient values,
The reception unit receives the coefficient value according to the corresponding unit width based on each unit width selected by the selection unit and the accepted parameter item.
The prediction device according to claim 3, characterized in that: A first correct answer rate indicating the rate at which the estimated ranking of each competition object matches the actual ranking without changing the value of each parameter item for past races; and the parameter item accepted by the reception unit. , and a second correct answer rate indicating a percentage of the estimated ranking of each sport object matching the actual ranking by changing the value of the target parameter item according to the coefficient value. Prepare,
The output unit further outputs comparison information comparing the first correct answer rate and the second correct answer rate calculated by the calculation unit.
The prediction device according to claim 3, characterized in that: A prediction method executed by a prediction device that predicts a race result for a race in which a plurality of competitive bodies move along a route in which a plurality of passing points including a goal are set,
a derivation step of inputting values for a plurality of parameter items regarding each competition object into a learned prediction model, and deriving a score value for each competition object for each passing point;
an estimating step of estimating the ranking of each competition object for each passing point based on each score value derived in the deriving step;
an output step of outputting prediction information including a change in the ranking of each of the sports objects in the form of an image or sound based on the ranking of each passing point estimated in the estimation step ;
The score value is a value that allows estimating the ranking of each competition object by sorting in descending order,
In the estimation step, the ranking of each of the competition objects is estimated for each passing point by sorting the score values of each of the competition objects in descending order,
The prediction model is a trained model that has been trained to output the score value for each of the sports objects for each passing point when values for a plurality of parameter items regarding each of the sports objects are input. ,
A prediction method characterized by : A computer that predicts race results for races in which multiple athletes move along routes with multiple passing points including the finish line.
a derivation unit that inputs values for a plurality of parameter items regarding each competition object into a learned prediction model, and derives a score value for each competition object for each passing point;
an estimating unit that estimates the ranking of each competition object for each passing point based on each score value derived by the deriving unit;
Functioning as an output unit that outputs prediction information including the transition of the ranking in each of the sports objects in the form of an image or sound based on the ranking for each passing point estimated by the estimating unit,
The score value is a value that allows estimating the ranking of each competition object by sorting in descending order,
The estimating unit estimates the ranking of each of the competition objects for each passing point by sorting the score values of each of the competition objects in descending order,
The prediction model is a trained model that has been trained to output the score value for each of the sports objects for each passing point when values for a plurality of parameter items regarding each of the sports objects are input. ,
A program characterized by :

Images