JP6948041B2 - Information processing equipment, information processing methods, and programs - Google Patents

Description

The present invention relates to a technique for predicting the price of real estate.

Patent Document 1 describes a technique for predicting the selling price when a property owned by a user is sold by using a prediction model. The forecast model is a model that learns the relationship between various property information such as property location, floor plan, size, age, presence or absence of remodeling, and the price at the time of sale.

Japanese Patent No. 6351779 (registered on June 15, 2018)

However, when predicting the price of a dwelling unit included in an apartment house by using the technique described in Patent Document 1, there are the following problems.

In apartment houses, new construction prices are set in order according to other conditions between multiple dwelling units with similar conditions, and it is expected that the selling prices will also be in that order. For example, among a plurality of dwelling units arranged one above the other with similar floor plans, the higher the floor, the higher the new construction price. In this case, the selling price of these multiple dwelling units is expected to be higher as the floor is higher. However, with the technology described in Patent Document 1, the prices predicted for a plurality of dwelling units with similar conditions in an apartment house are not necessarily in the order of the new construction price, and there is room for improvement from the viewpoint of accuracy. ..

One aspect of the present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide a technique for more accurately predicting the price of a dwelling unit included in an apartment house.

In order to solve the above-mentioned problems, the information processing apparatus according to one aspect of the present invention uses the input information acquisition unit for acquiring the input information for specifying the dwelling unit included in the condominium, and the input information using the prediction model. It is provided with a prediction unit for predicting the price of the dwelling unit at the first time point indicated by. The prediction model includes the price of the first dwelling unit included in the apartment house at the first time point and the price of the second dwelling unit similar to the first dwelling unit in the apartment house at the first time point. To maintain and predict the magnitude relationship between the price of the first dwelling unit at the second time point before the first time point and the price of the second dwelling unit at the second time point. Has been done.

According to one aspect of the present invention, the price of a dwelling unit included in an apartment house can be predicted more accurately.

It is a block diagram which shows the functional structure of the information processing system which concerns on one Embodiment of this invention. It is a figure which shows an example of the dwelling unit information in one Embodiment of this invention. It is a flowchart which shows the flow of the learning process executed by the information processing system 1 which concerns on one Embodiment of this invention. It is a figure explaining the detailed example of the process of calculating the average contract price and the ratio to a new construction price. It is a figure explaining the detailed example of the process which trains a prediction model. It is a figure which shows the specific example of the monotonicity constraint which is given when training a prediction model. It is a flowchart which shows the flow of the prediction processing executed by the information processing system 1 which concerns on one Embodiment of this invention. It is a figure explaining the detailed example of the process which outputs the predicted price using a forecast model. It is a block diagram which illustrates the physical configuration of the computer used as a server 10 and a user terminal 20.

[Embodiment]
Hereinafter, the information processing system 1 according to the embodiment of the present invention will be described in detail.

<Overview of Information Processing System 1>
The information processing system 1 predicts the price of the dwelling unit included in the apartment house at the first time point using the prediction model M1. Hereinafter, the dwelling unit for which the price is to be predicted is also described as the target dwelling unit. The first time point is the time point to be predicted. For example, the first time point is a tentative date on which it is assumed that a second-hand sale of the target dwelling unit will be concluded, but the date is not limited to this. Hereinafter, the first time point will also be described as the time to be predicted. In addition, the price at the first time point is also described as the predicted price.

<Outline of prediction model M1>
The prediction model M1 described above sets the price of the first dwelling unit at the first time point and the price of the second dwelling unit at the first time point as the price of the first dwelling unit at the second time point and the second time point. It is a trained model that has been machine-learned to maintain and predict the magnitude relationship with the price of the dwelling unit at the second time point.

Here, the first dwelling unit is an arbitrary dwelling unit included in the same apartment house as the target dwelling unit. The second dwelling unit is a dwelling unit similar to the first dwelling unit in an apartment house including the first dwelling unit. "Similar dwelling units" means that at least one of the attributes of the dwelling units included in the apartment house is similar. Such attributes include, but are not limited to, for example, area, orientation (direction in which the dwelling unit faces), floor plan, and the like. Similar areas may mean, for example, that the area difference is less than or equal to the threshold. Further, the similar orientation may mean that, for example, the classification when the dwelling units included in the apartment house are classified by orientation is the same. Further, the fact that the floor plans are similar may mean that, for example, the classification when the dwelling units included in the apartment house are classified by the floor plan is the same.

For example, in an apartment house, the dwelling unit arranged upward or downward of the first dwelling unit is likely to be a second dwelling unit similar to the first dwelling unit. This is because a plurality of dwelling units arranged in the vertical direction in an apartment house often have similar areas, orientations, floor plans, and the like. Therefore, if any one of the plurality of dwelling units arranged in the vertical direction is set as the first dwelling unit, there is a high possibility that the other one or more dwelling units are each the second dwelling unit.

The second time point is a time point before the first time point. The second time point may be any time point as long as the price of both the first dwelling unit and the second dwelling unit is determined at the second time point. For example, at the time of new construction, the price of both the first dwelling unit and the second dwelling unit at the second time point is fixed. In the present embodiment, it is assumed that the second time point is the time of new construction. Hereinafter, the second time point will also be referred to as the time of new construction. In addition, the price at the second time point is also described as the new construction price. In the present embodiment, the new construction price is the selling price at the time of new construction, but the contract price at the time of new construction may be used.

Further, the prediction model M1 is one or a plurality of dwelling units included in the condominium at one or a plurality of third time points after the second time point (at the time of new construction) and before the first time point (at the time of prediction target). The price of the first dwelling unit at the first time point (when predicted) and the price of the second dwelling unit at the first time point (when predicted) (forecast) with reference to the prices determined in Learned to predict price).

Here, "one or a plurality of third time points after the time of new construction and before the time of prediction target" is, for example, the time when a second-hand sale is concluded. Further, the price determined at such a third time point is the contract price of second-hand sales. Hereinafter, the third time point will also be referred to as the time when the contract is concluded. In addition, the price determined at the third time point is also described as the contract price.

For example, the prediction model M1 refers to one or more contract prices by inputting a feature amount calculated from one or more contract prices in an apartment house. In the present embodiment, the average value of one or more contract prices is applied as the feature amount calculated from one or more contract prices in the apartment house, and hereinafter, the feature amount is also described as the average contract price.

Further, the prediction model M1 refers to the comparative information in the condominium, and refers to the price (prediction price) of the first dwelling unit at the first time point (prediction target time) and the first time point (prediction target) of the second dwelling unit. Learned to predict the price (predicted price) at (hours). The comparative information in the apartment house was determined at the third time point (at the time of contract) closest to the first time point (predicted time) among one or more third time points (at the time of contract) in the apartment house. This is information comparing the price with the price at the second time point (at the time of new construction) of the dwelling unit for which the price was determined.

In the present embodiment, as the comparative information, the ratio of the contracted price to the new construction price for the most recently contracted dwelling unit in the condominium is applied, and thereafter, the comparative information is also described as the ratio to the new construction price.

Specifically, the prediction model M1 outputs the first predicted price at the time of the prediction target of the first dwelling unit by inputting the information indicating the predetermined feature amount of the first dwelling unit. Further, the prediction model M1 outputs the second predicted price at the time of the prediction target of the second dwelling unit by inputting the information indicating the predetermined feature amount of the second dwelling unit. At this time, the prediction model M1 is machine-learned so that the magnitude relationship between the first predicted price and the second predicted price becomes the same as the magnitude relationship between the new construction price of the first dwelling unit and the second dwelling unit. Examples of the predetermined feature amount to be input to the prediction model M1 include the dwelling unit information described later associated with the dwelling unit, the above-mentioned average contract price, the ratio to the new construction price, and the like. The details of the predetermined feature amount will be described later.

<Configuration of information processing system 1>
FIG. 1 is a block diagram showing a functional configuration of the information processing system 1 according to the embodiment of the present invention. In FIG. 1, the information processing system 1 includes a server 10 and a user terminal 20. The server 10 is an example of the information processing device in the present invention. The server 10 and the user terminal 20 are communicably connected via the network N1. The network N1 connecting the server 10 and the user terminal 20 is a wired LAN (Local Area Network), a wireless LAN, the Internet, a public network, a mobile data communication network, or a combination thereof.

(Configuration of server 10)
The server 10 includes a learning phase execution unit 11, a storage unit 12, and a prediction phase execution unit 13. The learning phase execution unit 11 includes a dwelling unit information acquisition unit 111 and a learning unit 112. The storage unit 12 stores the prediction model M1 and the dwelling unit information T1. The prediction phase execution unit 13 includes an input information acquisition unit 131 and a prediction unit 132.

The dwelling unit information acquisition unit 111 acquires dwelling unit information T1 for each of the plurality of dwelling units included in the plurality of apartment houses. Each dwelling unit information T1 includes a price (new construction price) of the dwelling unit at a second time point and a price (contract price) determined at one or a plurality of third time points (contract price) for the dwelling unit. The plurality of dwelling unit information T1 may be acquired from the outside or may be acquired from the storage unit 12. When acquired from the outside, the acquired plurality of dwelling unit information T1 is stored in the storage unit 12. The details of the dwelling unit information T1 will be described later.

The learning unit 112 trains the prediction model M1 using a plurality of dwelling unit information T1s as teacher data. The trained prediction model M1 is stored in the storage unit 12. As a learning algorithm for training the prediction model M1, a known learning algorithm can be applied. In this embodiment, machine learning (for example, LightGBM) using gradient boosting based on a decision tree algorithm is used. The outline of the prediction model M1 is as described above. The details of the learning unit 112 will be described later.

The input information acquisition unit 131 acquires input information that identifies the target dwelling unit included in the apartment house. The input information includes, for example, information for identifying an apartment house and information for specifying a dwelling unit in the apartment house.

The prediction unit 132 uses the prediction model M1 to predict the price (prediction price) of the target dwelling unit indicated by the input information at the first time point. Specifically, the prediction unit 132 outputs the predicted price of the target dwelling unit at the first time point by inputting the information indicating the predetermined feature amount of the target dwelling unit indicated by the input information into the prediction model M1. The details of the prediction process using the prediction model M1 will be described later.

Further, the prediction unit 132 uses information indicating the predicted price output from the prediction model M1 as a feature amount other than the predetermined feature amount of the target dwelling unit (that is, the predetermined feature amount input to the prediction model M1). It may be corrected. Details of the forecast price correction will be described later.

(Details of dwelling unit information T1)
FIG. 2 is a diagram showing an example of dwelling unit information T1 stored in the storage unit 12 of the server 10. Each row of the table shown in FIG. 2 indicates dwelling unit information T1. The dwelling unit information T1 includes room information, apartment house information, and contract information 1 to N (N is an integer of 1 or more).

The room information includes a dwelling unit ID that identifies the dwelling unit, a room number, an area, an orientation, a floor, and a new construction price. Among the information included in the room information, information other than the information used for identifying the dwelling unit (dwelling unit ID and room number) can be used as a predetermined feature amount input to the prediction model M1. Of the information included in the room information, the new construction price is an example of the "price at the second time point" in the present invention. In addition, among the information included in the room information, the area and the orientation are also used as attributes for determining the degree of similarity between the dwelling units. In addition, "direction" does not have to be information that strictly indicates the direction in which the dwelling unit is facing, but is information that indicates the classification (for example, facing south or east) when the dwelling units included in the apartment house are classified by direction. There may be. Hereinafter, a dwelling unit having a dwelling unit ID of Rxxx will also be referred to as a dwelling unit Rxxx.

The apartment house information includes the apartment house ID, name, location, age, and structure that identify the apartment house. Among the information included in the apartment house information, information other than the information used for identifying the apartment house (apartment house ID and name) can be used as a predetermined feature amount input to the prediction model M1. The location is information that identifies the physical location of the apartment building and is represented by, for example, latitude and longitude. The age of the building is information indicating the elapsed period from the time of new construction. The age of the building is not limited to the yearly unit as shown in the figure, but may be expressed in other units such as monthly units and daily units. The age of the building is updated at a predetermined timing (for example, a fixed time every day, a fixed date and time every month, a time point specified by input, etc.). The date of new construction may be used instead of the age of construction. The structure is information indicating the structure of a building (wooden structure, steel frame, reinforced concrete (RC), etc.) constituting an apartment house. Hereinafter, the apartment house whose apartment ID is Mxxx will also be referred to as an apartment house Mxxx.

Each of the contract information i (i is an integer of 1 or more and N or less) is represented by a set of the contract price i and the contract date i. Contract information 1 to N does not include information contracted at the time of new construction. In other words, the dwelling unit indicated by the dwelling unit information T1 has a record of contracts N times by second-hand sales after the time of new construction. Since the number of contracts concluded may differ depending on the dwelling unit, N is a value according to the dwelling unit. If there is no contract record (N = 0), the contract information is empty.

The contract date i included in the contract information i indicates the time when the second-hand sale of the dwelling unit is concluded. In other words, the closing date i is an example of the third time point in the present invention. The contract price i indicates the contract price on the contract date i. Part or all of the contract information 1 to N is used to calculate the average contract price and the ratio of the new construction price to the input to the prediction model M1. In addition, a part or all of the contract information 1 to N is also used as a correct answer of the output from the prediction model M1.

The dwelling unit information T1 does not necessarily have to be stored in the storage unit 12 as a table having the structure shown in FIG. For example, the dwelling unit information T1 includes a room information table having a dwelling unit ID as a primary key and a condominium ID as a foreign key, a condominium information table having a condominium ID as a primary key, and a contract information table using a dwelling unit ID as a foreign key. It may be composed of. Further, not limited to this, the dwelling unit information T1 is stored in another data structure as long as it is a data structure capable of specifying the above-mentioned room information, condominium information and contract information for the dwelling unit based on the dwelling unit ID. You may.

(Configuration of user terminal 20)
As shown in FIG. 1, the user terminal 20 includes a control unit 21 and a storage unit 22. The control unit 21 transmits the input information for specifying the target dwelling unit input by the user to the server 10. Further, the control unit 21 outputs the estimated price of the target dwelling unit transmitted from the server 10. Information transmitted to and received from the server 10 is stored in the storage unit 22.

<Learning process of information processing system 1>
Next, the flow of the learning process of the information processing system 1 will be described.

FIG. 3 is a flowchart showing the flow of the learning process executed by the server 10.

In step S101, the dwelling unit information acquisition unit 111 acquires dwelling unit information T1 for each of the plurality of dwelling units included in the plurality of apartment houses. The dwelling unit information acquisition unit 111 may acquire these dwelling unit information T1 stored in the storage unit 12, or may acquire these dwelling unit information T1 from the outside. When acquired from the outside, these acquired dwelling unit information T1 are stored in the storage unit 12.

In step S102, the learning unit 112 refers to the acquired plurality of dwelling unit information T1 and calculates the average contract price and the ratio to the new construction price. The details of the step will be described later.

In step S103, the learning unit 112 trains the prediction model M1 using the teacher data composed of the dwelling unit information T1 acquired in step S101, the average contract price calculated in step S102, and the new construction price ratio. The details of the step will be described later. Further, the learning unit 112 stores the learned prediction model M1 in the storage unit 12.

With the above, the server 10 ends the learning process.

(Details of calculation process of average contract price and new construction price ratio)
FIG. 4 is a diagram illustrating a detailed example of the process of calculating the average contract price and the new construction price ratio in step S102. Both the average contract price and the ratio of the new construction price to the new construction price are calculated for each apartment house M according to the time point t of interest. The time of interest t is an arbitrary time after the time of new construction. In other words, the average contract price and the ratio of the new construction price to the new construction price are parameters determined according to the apartment house ID and the time point t of interest.

The average contract price is the average of the contract prices determined after the time of new construction and before the time of interest t in the apartment house M. In the example shown in FIG. 4, the average contract price of the apartment house M001 at the time of attention in November 2011 was 51 million, which is a value calculated based on the example of FIG. In the example of FIG. 2, in the condominium M001, before the time of attention in November 2019, the second-hand sales of the dwelling unit R001 twice and the dwelling unit R002 once, a total of three times, have been concluded. The contract prices at the time of these three contracts are 50 million, 48 million, and 55 million. Therefore, these average 51 million are calculated as the average contract price.

Further, in the example shown in FIG. 4, the average contract price of the apartment house M001 at the time of attention in October 2019 was 52.5 million, which is a value calculated based on the example of FIG. In the example of FIG. 2, in the condominium M001, before October 2019, the second-hand sales of the dwelling unit R001 and once for the dwelling unit R002 have been concluded, for a total of two times. The contract prices at the time of these two contracts are 50 million and 55 million. Therefore, these averages of 52.5 million are calculated as the average contract price.

The ratio to the new construction price is the ratio of the latest contract price before the time of interest t in the apartment house M to the new construction price of the dwelling unit for which the contract price has been determined. In the example shown in FIG. 4, the ratio of the new construction price to the new construction price as of November 2011 at the time of attention of the apartment house M001 is 0.80, which is a value calculated based on the example of FIG. In the example of FIG. 2, in the apartment house M001, the latest contract price before November 2019 at the time of attention is 48 million, which was determined for the dwelling unit R001 in October 2019. Therefore, as the ratio to the new construction price, the above-mentioned 0.80 is calculated as the ratio of the contract price of 48 million in October 2019 to the new construction price of 60 million of the dwelling unit R001.

Further, in the example shown in FIG. 4, the ratio of the new construction price to the new construction price as of October 2011 at the time of attention of the apartment house M001 is 0.83, which is a value calculated based on the example of FIG. In the example of FIG. 2, in the apartment house M001, the latest contract price before October 2019 at the time of attention is 50 million, which was determined for the dwelling unit R001 in January 2015. Therefore, as the ratio to the new construction price, the above-mentioned 0.83 is calculated as the ratio of the contract price of 50 million in January 2015 to the new construction price of 60 million of the dwelling unit R001.

(Details of learning process)
FIG. 5 is a diagram illustrating a detailed example of the process of training the prediction model M1 in step S103. As shown in FIG. 5, the learning unit 112 includes information (1) to (4) excluding the contract price n and contract information (n + 1) to N from the dwelling unit information T1 of a dwelling unit, and a condominium including the dwelling unit. The prediction model M1 is trained so as to output (7) the contract price n by inputting the information (5) to (6) calculated with the contract date n as the time of interest. n is an integer of 1 or more and N or less. Specifically, the learning unit 112 includes (1) room information (other than the dwelling unit ID and room number), (2) condominium information (other than the common dwelling unit ID and name), and (3) contract information 1 among the dwelling unit information T1. -(N-1), (4) Contract date n, (5) Average contract price, and (6) To new construction price ratio are input, and (7) Contract price n is trained to be output. .. The (5) average contract price and (6) ratio of new construction price to the new construction price are information calculated with the contract date n as the time of interest for the condominium including the dwelling unit.

In other words, the dwelling unit information T1 is input to the prediction model M1. Information 1 to (n-1), (4) Contract date n, (5) Average contract price, and (6) New construction price ratio ", and" (7) Contract price, which is the correct answer for the output from the prediction model M1. It is used as teacher data associated with "n".

In addition, the dwelling unit information T1 includes N contract information. By applying an integer of 1 or more and N or less as n as described above, the dwelling unit information T1 can be used as teacher data including a maximum of N sets of input and output (correct answers). For example, the learning unit 112 may use the dwelling unit information T1 as teacher data by applying each contract price whose contract date is after the threshold value as the contract price n that is the correct answer. Further, the learning unit 112 may use the dwelling unit information T1 as teacher data by applying a predetermined number of contract prices in order of newest contract date as the contract price n that is the correct answer.

Further, the learning unit 112 applies a monotonicity constraint to train the prediction model M1. The monotonicity constraint is a constraint in machine learning that causes the output to be monotonically increased (or monotonically decreased) with respect to an increase in input features. Here, the new construction price included in the room information is applied as a feature amount that gives a monotonic constraint that monotonically increases the output. In addition, as a range to realize the monotonic constraint, the range of similar dwelling units is applied in the same apartment house.

(Specific example of monotonicity constraint)
FIG. 6 is a diagram showing a specific example of the monotonicity constraint. The apartment house in this example has five dwelling units on each floor from the first floor to the fifth floor. In the three-digit room number assigned to each dwelling unit, the hundreds digit represents the floor, and the ones digit is the same among the plurality of dwelling units arranged one above the other. In addition, it is assumed that the plurality of dwelling units arranged one above the other have similar areas and orientations, and that the higher the floor, the higher the new construction price is set between these dwelling units. Specifically, the new construction prices of room numbers 10x, 20x, 30x, 40x, 50x (x = 1, 2, 3, 4, 5) increase in this order. In this case, the learning unit 112 applies the monotonicity constraint that the output is monotonically increased with respect to the increase in the new construction price between similar dwelling units in the learning of the prediction model M1. As a result, the predicted price output from the predicted model M1 becomes higher as the new construction price is higher.

As described above, the similarity between dwelling units may be determined based on the similarity in area and orientation, or may be determined based on the room number.

<Prediction processing of information processing system 1>
Next, the flow of the prediction process of the information processing system 1 will be described.

FIG. 7 is a flowchart showing a flow of prediction processing executed by the server 10 and the user terminal 20.

In step S201, the control unit 21 of the user terminal 20 acquires the input information for identifying the target dwelling unit via the input device. Further, the control unit 21 transmits the acquired input information to the server 10. For example, the input information may include the location and name of the apartment house including the target dwelling unit and the room number of the target dwelling unit. The location may be part or all of the address. Further, the input information may include information indicating the time to be predicted. If the input information does not include information indicating the prediction target time, a time point satisfying a predetermined condition (for example, an execution time point of the step) is set as the prediction target time.

In step S202, the input information acquisition unit 131 of the server 10 acquires the above-mentioned input information by receiving it from the user terminal 20.

In step S203, the prediction unit 132 identifies the dwelling unit information T1 of the target dwelling unit indicated by the input information among the dwelling unit information T1 stored in the storage unit 12. For example, the prediction unit 132 extracts the dwelling unit information T1 in the area where the position of the apartment house is calculated from the location included in the input information from the dwelling unit information T1 stored in the storage unit 12. Further, the prediction unit 132 further extracts the dwelling unit information T1 in which the name of the apartment house matches the name included in the input information from the extracted dwelling unit information T1. Further, the prediction unit 132 identifies the dwelling unit information T1 whose room number matches the room number included in the input information among the extracted dwelling unit information T1. The configuration of the input information and the process of identifying the dwelling unit information T1 from the input information are not limited to those described above.

In step S204, the prediction unit 132 calculates the average contract price and the ratio of the new construction price to the new construction price at the time of prediction for the apartment house including the target dwelling unit with reference to the dwelling unit information T1 of the specified target dwelling unit. The details of the step will be similarly described by applying the time to be predicted as the time of interest t in the calculation process described with reference to FIG.

In step S205, the prediction unit 132 outputs the predicted price of the target dwelling unit at the time of the prediction target using the prediction model M1 stored in the storage unit 12. The estimated price is transmitted to the user terminal 20. The details of the step will be described later.

In step S206, the control unit 21 of the user terminal 20 causes a display device or the like to display the received estimated price.

With the above, the server 10 ends the prediction process.

(Details of prediction processing)
FIG. 8 is a diagram illustrating a detailed example of the process of outputting the predicted price of the target dwelling unit using the prediction model M1 in step S205. As shown in FIG. 8, the prediction unit 132 includes (1) room information (other than the dwelling unit ID and room number) and (2) condominium information (other than the communal dwelling unit ID and name) included in the dwelling unit information T1 specified in step S203. ), And (3) Contract information 1 to N, (5) Average contract price calculated in step S204, (6) Ratio of new construction price, and (8) Prediction target time included in the input information. Enter in M1. These pieces of information input to the prediction model M1 are examples of information indicating a predetermined feature amount in the present invention. As a result, (9) the predicted price at the time of the prediction target of the target dwelling unit is output from the prediction model M1.

Further, in step S205, the prediction unit 132 sets the predicted price obtained by inputting the information indicating the predetermined feature amount of the target dwelling unit into the prediction model M1 and the feature amount other than the predetermined feature amount of the target dwelling unit. The information shown may be used for correction. Since the example of the predetermined feature amount is as described above, the detailed description will not be repeated. In this case, as an example of the feature amount other than the predetermined feature amount, for example, there is information indicating that the property is an accident property, information indicating the addition of equipment, and the like. As an example, when the target dwelling unit is an accident property, the prediction unit 132 may correct the predicted price output from the prediction model M1 to a lower value. Further, when the target dwelling unit has equipment (air conditioner, floor heating, etc.) that the target dwelling unit does not have at the time of new construction, the prediction unit 132 may correct the predicted price output from the prediction model M1 to a higher value.

<Effect of this embodiment>
In the present embodiment, among similar first dwelling units and second dwelling units included in the same condominium, by inputting information regarding the first dwelling unit in step S201, the first predicted price is set in step S206. It is output. Further, by inputting the information regarding the second dwelling unit in step S201, the second predicted price is output in step S206. In this embodiment, by using the prediction model M1, if the prediction target times are the same, the magnitude relationship between the first predicted price and the second predicted price, and the new construction of the first dwelling unit and the second dwelling unit. The size relationship between prices is the same. As a result, in the present embodiment, the predicted price of the target dwelling unit at the time of the forecast is more accurately predicted by reflecting the magnitude relationship of the new construction price with the dwelling unit similar to the target dwelling unit in the condominium. Can be done.

[Modification example]
<Variation of configuration>
In the above-described embodiment, the server 10 has been described as including the learning phase execution unit 11 and the prediction phase execution unit 13. However, each of these parts may be configured as a separate device.

Further, the server 10 does not have to include the prediction model M1. In this case, by connecting the server 10 to an external device that stores the prediction model M1, the server 10 operates in the same manner as in the above-described embodiment and has the same effect.

Further, the server 10 may be configured to include the prediction phase execution unit 13 and not the learning phase execution unit 11. In this case, the server 10 can more accurately predict the price of the target dwelling unit included in the apartment house at the first time point by using the prediction model M1.

Further, the server 10 may be configured to include the learning phase execution unit 11 and not the prediction phase execution unit 13. In this case, the server 10 can generate a prediction model M1 for more accurately predicting the price of the target dwelling unit included in the apartment house at the first time point.

<Variations of learning process>
Further, in the above-described embodiment, the prediction model M1 has been described as being trained using LightGBM. However, the learning algorithm for training the prediction model M1 is not limited to this, and other known learning algorithms can be applied. For example, the prediction model M1 may be a model learned by other machine learning (support vector machine, etc.), deep learning (CNN: Convolutional Neural Networks, RNN: Recurrent Neural Network), GAN: Generative Adversarial Network, etc.). good.

<Other variations>
Further, in the above-described embodiment, as an example of a plurality of similar dwelling units, dwelling units arranged in the vertical direction are mentioned, but the present invention is not limited to this. For example, a plurality of similar dwelling units may be dwelling units included in the same building, dwelling units included in the same floor, or a plurality of dwelling units having other similarities.

[Example of realization by software]
The control block of the server 10 (particularly, the dwelling unit information acquisition unit 111, the learning unit 112, the input information acquisition unit 131, and the prediction unit 132) is realized by a logic circuit (hardware) formed in an integrated circuit (IC chip) or the like. It may be realized by software. In the latter case, each of the server 10 and the user terminal 20 is configured by using, for example, a computer (electronic computer).

(Physical configuration of server 10)
FIG. 9 is a block diagram illustrating the physical configuration of a computer used as the server 10 and the user terminal 20. As shown in FIG. 9, the server 10 can be configured by a computer including a bus 110, a processor 101, a main memory 102, an auxiliary memory 103, and a communication interface 104. The processor 101, the main memory 102, the auxiliary memory 103, and the communication interface 104 are connected to each other via the bus 110.

As the processor 101, for example, a CPU (Central Processing Unit), a microprocessor, a digital signal processor, a microcontroller, or a combination thereof or the like is used.

As the main memory 102, for example, a semiconductor RAM (random access memory) or the like is used.

As the auxiliary memory 103, for example, a flash memory, an HDD (Hard Disk Drive), an SSD (Solid State Drive), or a combination thereof or the like is used. The auxiliary memory 103 stores a program for causing the processor 101 to execute the operation of the server 10 described above. The processor 101 expands the program stored in the auxiliary memory 103 on the main memory 102, and executes each instruction included in the expanded program.

The communication interface 104 is an interface that connects to the network N1.

In this example, the processor 101 and the communication interface 104 are examples of hardware elements that realize the dwelling unit information acquisition unit 111, the learning unit 112, the input information acquisition unit 131, and the prediction unit 132. Further, the main memory 102 and the auxiliary memory 103 are examples of hardware elements that realize the storage unit 12.

(Physical configuration of user terminal 20)
As shown in FIG. 9, the user terminal 20 can be configured by a computer including a bus 210, a processor 201, a main memory 202, an auxiliary memory 203, a communication interface 204, and an input / output interface 205. The processor 201, the main memory 202, the auxiliary memory 203, the communication interface 204, and the input / output interface 205 are connected to each other via the bus 210. An input device 206 and an output device 207 are connected to the input / output interface 205.

As the processor 201, for example, a CPU, a microprocessor, a digital signal processor, a microcontroller, or a combination thereof or the like is used.

As the main memory 202, for example, a semiconductor RAM or the like is used.

As the auxiliary memory 203, for example, a flash memory, an HDD, an SSD, or a combination thereof or the like is used. The auxiliary memory 203 stores a program for operating the computer as a user terminal 20. The processor 201 expands the program stored in the auxiliary memory 203 on the main memory 202, and executes each instruction included in the expanded program. Further, the auxiliary memory 203 stores various data referred to by the processor 201 in order to operate the computer as the user terminal 20.

The communication interface 204 is an interface that connects to the network N1.

As the input / output interface 205, for example, a USB interface, a short-range communication interface such as infrared rays or Bluetooth (registered trademark), or a combination thereof is used.

As the input device 206, for example, a keyboard, a mouse, a touch pad, a microphone, or a combination thereof or the like is used. As the output device 207, for example, a display, a printer, a speaker, or a combination thereof is used.

In this example, the processor 201 and the communication interface 204 are examples of hardware elements that realize the control unit 21. Further, the main memory 202 and the auxiliary memory 203 are examples of hardware elements that realize the storage unit 22.

The above-mentioned programs may be recorded in an external recording medium instead of being stored in the auxiliary memories 103 and 203, respectively, and may be supplied to the corresponding computer by being read from the external recording medium. As the external recording medium, a computer-readable "non-temporary tangible medium" such as a tape, a disk, a card, a semiconductor memory, a programmable logic circuit, or the like can be used. Further, each of the above-mentioned programs may be supplied to the computer via an arbitrary transmission medium (communication network, broadcast wave, etc.) that can be transmitted. In addition, one aspect of the present invention can also be realized in the form of a data signal embedded in a carrier wave, in which each program is embodied by electronic transmission.

The present invention is not limited to the above-described embodiments, and various modifications can be made within the scope of the claims, and the embodiments obtained by appropriately combining the technical means disclosed in the different embodiments. Is also included in the technical scope of the present invention.

〔summary〕
In order to solve the above problems, the information processing apparatus according to one aspect of the present invention uses the input information acquisition unit for acquiring the input information for specifying the dwelling unit included in the condominium, and the input information using the prediction model. The prediction unit includes a prediction unit for predicting the price of the dwelling unit at the first time point indicated by the above, and the prediction model includes the price of the first dwelling unit included in the condominium at the first time point and the said in the condominium. The price of the second dwelling unit similar to the first dwelling unit at the first time point, the price of the first dwelling unit at the second time point before the first time point, and the second dwelling unit. It is learned to maintain and predict the magnitude relationship with the price at the second time point.

According to the above configuration, the predicted price of the dwelling unit included in the apartment house reflects the magnitude relationship of the price at the second time point between similar dwelling units in the apartment house. As a result, the price of the dwelling unit included in the apartment house can be predicted more accurately.

In the information processing apparatus according to one aspect of the present invention, the prediction model is included in the condominium at one or a plurality of third time points after the second time point and before the first time point. Alternatively, it is learned to predict the price of the first dwelling unit at the first time point and the price of the second dwelling unit at the first time point with reference to the prices determined for each of the plurality of dwelling units. Is preferable.

According to the above configuration, the estimated price of the dwelling unit included in the apartment house was determined at one or more third time points after the second time point and before the first time point in the apartment house. The price will be reflected. As a result, the price of the dwelling unit included in the apartment house can be predicted more accurately.

In the information processing apparatus according to one aspect of the present invention, the prediction model is a price determined at a third time point closest to the first time point among the one or a plurality of third time points, and the price. With reference to the comparative information comparing the price of the determined dwelling unit with the price at the second time point, the price of the first dwelling unit at the first time point and the price of the second dwelling unit at the first time point. It is preferable that it is learned to predict the price in.

According to the above configuration, the estimated price of the dwelling unit included in the apartment house is the price determined at the third time point closest to the first time point in the apartment house and the price at the second time point of the dwelling unit. The comparison information with the price will be reflected. As a result, the price of the dwelling unit included in the apartment house can be predicted more accurately.

In the information processing device according to one aspect of the present invention, the prediction unit inputs information indicating a predetermined feature amount of the dwelling unit indicated by the input information into the prediction model, so that the price of the dwelling unit at the first time point. Is output, and the output price is preferably corrected by using information indicating a feature amount other than the predetermined feature amount of the dwelling unit.

According to the above configuration, the effect of the feature quantity not included in the input to the prediction model is taken into consideration on the price of the dwelling unit included in the apartment house, so that the price of the dwelling unit at the first time point is predicted more accurately. can do.

It is preferable that the information processing apparatus according to one aspect of the present invention further includes a learning unit for learning the prediction model.

According to the above configuration, it is possible to train the prediction model for the above-mentioned information processing apparatus to exert the above-mentioned effect.

In order to solve the above problems, the information processing device according to one aspect of the present invention is similar to the price of the first dwelling unit included in the apartment house at the first time point and the first dwelling unit in the apartment house. The price of the second dwelling unit at the first time point, the price of the first dwelling unit at the second time point before the first time point, and the price of the second dwelling unit at the second time point. It is provided with a learning unit that trains a prediction model so as to maintain and predict the magnitude relationship with the price in.

According to the above configuration, it is possible to learn a prediction model for more accurately predicting the price of a dwelling unit included in an apartment house.

In order to solve the above problems, the information processing method according to one aspect of the present invention uses an input information acquisition step in which the information processing device acquires input information for identifying a dwelling unit included in a condominium, and a prediction model. It is an information processing method that executes a prediction step of predicting the price of the dwelling unit at the first time point indicated by the input information, and the prediction model is the first of the first dwelling units included in the condominium. The price at the time of 1 and the price of the second dwelling unit similar to the first dwelling unit in the condominium at the first time point are set to the first of the first dwelling unit before the first time point. It is learned to maintain and predict the magnitude relationship between the price at the second time point and the price of the second dwelling unit at the second time point.

According to the above configuration, the same effect as that of the above-mentioned information processing apparatus is obtained.

In order to solve the above problems, the program according to one aspect of the present invention is a program for operating a computer as the above-mentioned information processing device, and causes the computer to function as each of the above-mentioned parts.

According to the above configuration, the same effect as that of the above-mentioned information processing apparatus is obtained.

1 Information processing system 10 Server 20 User terminal 11 Learning phase execution unit 12, 22 Storage unit 13 Prediction phase execution unit 21 Control unit 111 Dwelling unit information acquisition unit 112 Learning unit 131 Input information acquisition unit 132 Prediction unit 101, 201 Processor 102, 202 Main memory 103, 203 Auxiliary memory 104, 204 Communication interface 110, 210 Bus 205 Input / output interface 206 Input device 207 Output device

Claims (8)

The input information acquisition unit that acquires the input information that identifies the dwelling unit included in the apartment house,
Using the forecast model, the forecast unit that predicts the forecast price at the time of the forecast target when the contract for the second-hand sales of the dwelling unit indicated by the input information is assumed,
With
The prediction model is
The predicted price of the first dwelling unit included in the apartment house and the predicted price of the second dwelling unit similar to the first dwelling unit in the apartment house.
With reference to the apartment house information indicating the apartment house and the contract price of the second-hand sale of each of the first dwelling unit and the second dwelling unit before the forecast target time , the new construction price of the first dwelling unit and the new construction price of the first dwelling unit. , Learned to maintain and predict the magnitude relationship with the new construction price of the second dwelling unit,
An information processing device characterized by this. The forecast model refers to the contracted price of the second-hand sales contracted from the time of new construction of one or more dwelling units included in the condominium to the time of the forecast target, and the predicted price of the first dwelling unit and the predicted price. Learned to predict the forecast price of the second dwelling unit,
The information processing apparatus according to claim 1. The forecast model refers to comparative information comparing the contract price at the time of the contract closest to the forecast target and the new construction price of the dwelling unit for which the contract price has been determined among the contract prices of the one or more dwelling units. Then, it is learned to predict the predicted price of the first dwelling unit and the predicted price of the second dwelling unit.
The information processing apparatus according to claim 2. The prediction unit outputs the predicted price of the dwelling unit by inputting information indicating a predetermined feature amount of the dwelling unit indicated by the input information into the prediction model, and outputs the output price to the predetermined feature of the dwelling unit. Correction using information indicating feature quantities other than quantities,
The information processing apparatus according to any one of claims 1 to 3, wherein the information processing device is characterized by the above. A learning unit for learning the prediction model is further provided.
The information processing apparatus according to any one of claims 1 to 4. The predicted price at the time of the forecast when the contract for the second-hand sales of the first dwelling unit included in the condominium is assumed and the predicted price of the second dwelling unit similar to the first dwelling unit in the condominium are set.
With reference to the condominium information indicating the condominium and the contracted price of the pre-owned sale of each of the first dwelling unit and the second dwelling unit before the time of the forecast , the new construction price of the first dwelling unit and the new construction price of the first dwelling unit. An information processing device comprising a learning unit for learning a prediction model so as to maintain and predict a magnitude relationship with the new construction price of the second dwelling unit. Information processing device
Input information acquisition step to acquire input information to identify the dwelling unit included in the apartment house,
It is an information processing method that executes a prediction step of predicting the predicted price at the time of prediction when a contract for second-hand sales of a dwelling unit is assumed, which is indicated by the input information, using a prediction model.
The prediction model is
The predicted price of the first dwelling unit included in the apartment house and the predicted price of the second dwelling unit similar to the first dwelling unit in the apartment house.
With reference to the apartment house information indicating the apartment house and the contract price of the second-hand sale of each of the first dwelling unit and the second dwelling unit before the forecast target time , the new construction price of the first dwelling unit and the new construction price of the first dwelling unit. , Learned to maintain and predict the magnitude relationship with the new construction price of the second dwelling unit,
An information processing method characterized by the fact that. A program for operating a computer as the information processing device according to any one of claims 1 to 6, and for operating the computer as each part.

Images