JP2003022314A - Method, device and program for estimating real estate price function - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for estimating a price function, that estimates an optimum function with which the price of real estate is predicted in each divided segment by performing the market division corresponding to a complicated market structure. SOLUTION: This cost function estimating method for calculating a predicted price of a piece of real estate with a statistical analysis has a step for setting segments, in which a market structure is composed of homogeneous object data by performing a structural change test of the object data, a step for selecting a statistical model to be used to estimate the cost function from among a plurality of statistical models in each set segment, and a step for using the selected statistical model to estimate the cost function for the object data in each set segment.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for estimating a function for calculating a real estate price and an apparatus for estimating a function for calculating a real estate price.

[0002]

2. Description of the Related Art Conventionally, prices of real estate such as condominiums, houses, and land have been calculated by the following method. For example, if a so-called real estate appraiser does
The price was calculated by extracting a property similar to the property subject to evaluation from the real estate properties existing in the neighborhood of the property subject to evaluation and comparing this with the property subject to evaluation.

When a real estate price is calculated by constructing a predetermined system, the market to be analyzed is divided into a plurality of segments with reference to the line price of the property, and each segment is divided. Real estate prices were calculated by performing statistical analysis.

[0004]

However, in the above-mentioned price calculation method by the real estate appraiser, there are many parts based on the experience and intuition of each individual in the work of calculating the price.
There was a problem that reproducibility was difficult and lacking in objectivity. Furthermore, the person who requested the evaluation needs to judge whether the calculated price is valid or not, but there are many uncertainties in the price calculation work based on experience and intuition, and only the calculated price is available. It was difficult to judge the appropriateness of the price.

In addition, in the price calculation method using a predetermined system, the price was calculated for each market, but since the market to be analyzed was divided by a simple method, the market for an originally complicated real estate property was The structure could not be accommodated. Therefore, there is a problem in that the calculated price accuracy varies even within the same market, and for example, when the relationship between the distance from the station and the price is different across the line, an appropriate price cannot be predicted. there were.

[0006] Therefore, a first object of the present invention is to provide a price function estimation method for performing market division corresponding to a complicated market structure and estimating an optimal function for predicting the price of real estate for each divided segment. To provide.

Further, a second object of the present invention is a real estate price predicting apparatus capable of presenting information useful to a client when determining a price calculated by using a method of estimating a price function of real estate. To provide.

[0008]

A method for estimating a function of a real estate price according to the present invention performs a structural change test in consideration of various factors influencing the formation of a real estate price, so that a segment with the same market structure is identified. It is characterized in that an optimal price function is estimated by setting and selecting a statistical model or variable used for function estimation for each of the set segments.

That is, by performing a structural change test on the data relating to the price of the real estate property under a predetermined condition, segments with the same market structure in the real estate property are set, and the data of each set segment is targeted. In addition, the price function is estimated by selecting models and variables using various statistical methods.

As a result, it is possible to set an optimum segment that adapts to the real estate market structure, and it is possible to select models and variables that reflect the real estate market structure, and as a result, to estimate the optimal price function. You will be able to.

Structural change tests
t) means the equivalence test of regression coefficients. For example, the linear model formula y = a0 + b1x1 + b2x2 ... has the meaning of mathematically and statistically reproducing the condominium market, and the structure of the condominium market can be structurally estimated by this formula. Specifically, the structural change test is to test whether the two estimated parameters (b1, b2) are the same for the A market and the B market which are arbitrarily divided.

The market structure is expressed by the above (formula) y = a0 + b1x1 + b2x2
If the values of the parameters (b1, b2) are the same, the market structure is said to be homogeneous.

[0013] It is desirable to set the predetermined condition as an element that affects the market structure of the real estate property. For example, the behavior of consumers and suppliers that make up the market changes moment by moment depending on the content of goods, the flow of time, and the region. Therefore, in the present invention, as the conditions of the structural change test, real estate type, time,
And it is desirable to set the area.

[0014] The data on the price of real estate property includes
Any data that is considered to affect the price of the property is applicable, for example, address data, property attribute data representing the property of the property, and factor data created based on the regional information data.

The statistical model includes, for example, least squares method, maximum likelihood method, neural network, covariance structure analysis, GM.
M (Generalized Methods of Moment) and Spatial S
There are tatistics, etc., and linear models and non-linear models are targeted. In the present invention, the optimal market segment is set on the premise of the linear model, but in order to further improve the accuracy of the price function, the model with the highest prediction accuracy including the nonlinear model is selected.

That is, the price function estimating method according to the present invention is a method of estimating a price function of a real estate property by statistical analysis, in which a structural change test is performed on data relating to the price of the real estate property. A step of setting a segment composed of data on prices of real estate properties with the same market structure, and a step of selecting a statistical model used for estimating the price function from a plurality of statistical models for each of the set segments. Estimating the price function for each of the set segments using the selected statistical model for data relating to the price of the real estate property.

The step of setting the segment includes
Setting a first segment composed of price data having the same structure for the real estate type by performing a structural change test on the data related to the price of the real estate property on the condition of the real estate type; Performing a structural change test on the data on the price of the real estate property to which it belongs, by setting a time, to set a second segment consisting of data on the price of the real estate property with the same structure for the real estate type and time; By performing a structural change test on the data regarding the price of the real estate property belonging to the second segment based on the region, the third data consisting of the data regarding the price of the real estate property with the same structure for real estate type, time, and region Setting a segment, the step of estimating the price function comprising: The using of the selected statistical model to the data relating to prices belonging real estate to the segment and estimates the price function, the real estate type, it is desirable that the scale of the real estate.

The step of selecting the statistical model includes
A step of creating n test price functions by using n statistical models for the data related to the price of the real estate property, the transaction price of the real property property in the data related to the price of the real property property and the n created values Calculating a divergence rate based on an estimated price estimated by the test price function, and calculating a Fitting Rate representing the prediction accuracy of the created test price function from the calculated divergence rate,
Selecting the statistical model with the highest prediction accuracy from the n statistical models based on the calculated Fitting Rate.

Further, the price function estimating method includes the step of selecting an optimum variable converting method used for estimating the price function from a plurality of variable converting methods for each of the set segments. For each segment, estimating the price function by using the selected statistical model and the selected variable conversion method with respect to the data regarding the price of the real estate property, the method further comprising: , The method of estimating the price function, the step of inputting the address data of the real estate property and the property attribute data indicating the attributes of the property, and the step of identifying the coordinate data of the real estate property based on the input address data, Creating factor data for forming the price of the real estate property based on the specified coordinate data and a database storing regional information. To have, wherein the data relating to the price of real estate, the address data, the property attribute data, the identified coordinate data, and wherein said contain factor data created.

The method of estimating the price function further comprises the step of inputting data on the price of the real estate property, and estimates the price function according to the frequency of inputting the data on the price of the real estate property. It is characterized by doing.

Further, the price function estimation device according to the present invention is
In the price function estimation device that calculates the predicted price of real estate properties by statistical analysis, a structural change test is performed on the data related to the prices of real estate properties to determine the segment consisting of the data related to the prices of real estate properties with the same market structure. Means for setting, means for selecting a statistical model used for estimating the price function from a plurality of statistical models for each of the set segments, and data regarding the price of the real estate property for each of the set segments And means for estimating the price function using the selected statistical model.

Further, the program according to the present invention is a price function estimation program for calculating a predicted price of a real estate property by statistical analysis, and by performing a structural change test on data relating to the price of the real estate property, A function of setting a segment composed of data on prices of real estate properties having the same structure, a function of selecting a statistical model used for estimating the price function from a plurality of statistical models for each of the set segments, and The computer is made to realize the function of estimating the price function using the selected statistical model for the data regarding the price of the real estate property for each set segment.

Further, the real estate price forecasting apparatus according to the present invention performs a structural change test on the data relating to the price of the real estate property to set a segment consisting of the data relating to the price of the real estate property having the same market structure, A statistical model used for estimating the price function is selected from a plurality of statistical models for each of the set segments, and the selected statistical model is used for the data regarding the price of the real estate property for each of the set segments. Estimate the price function using, storage means for storing the estimated price function, input means for inputting the address data and property attribute data of the property to be evaluated, based on the input address data Based on the means for specifying the coordinate data of the property to be evaluated, based on the input property attribute data and the specified coordinate data Based on a database that stores the specified coordinate data and area information, specifying the segment to which the property to be evaluated belongs, reading the price function corresponding to the specified segment from the storage means, A means for calculating the living environment of the evaluation target property as an index, and a predicted price of the evaluation target property using the input property attribute data, the calculated index, and the read price function. Means and are provided.

Further, the real estate management system according to the present invention is a price function estimation program for calculating a predicted price of a real estate property by statistical analysis, and by performing a structural change test on data relating to the price of the real estate property. A function of setting a segment composed of data on prices of real estate properties with the same market structure, and a function of selecting a statistical model used for estimating the price function from a plurality of statistical models for each of the set segments. , A function of estimating the price function by using the selected statistical model for data regarding the price of the real estate property for each of the set segments, the price function of a computer. A method of inputting property information for identifying a collateral property or a property for sale, which has an estimation program. Means for calculating the price of the collateral property or the sale property by using the property information and the price function calculated by executing the estimation program of the price function, and the calculated price and the loan condition or sale Means for calculating a loan amount or a sale amount of the collateral property or the sale property based on the condition.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION [First Embodiment] Next, a first embodiment of the present invention (hereinafter referred to as "the present embodiment") will be described with reference to the drawings. In the first embodiment, a segment having the same market structure is identified by performing a structural change test on real estate data under conditions of real estate type, time, and area, and function estimation is performed based on the identified segment. A real estate price function estimating device for estimating an optimal real estate price function by selecting a model and a variable used for will be described. (Schematic Configuration of Real Estate Price Function Estimating Device) FIG. 1 is a block diagram showing a functional configuration of the real estate price function estimating device according to the present embodiment. The real estate price function estimating device (hereinafter referred to as “this device”) according to the present embodiment includes real estate price data input means 1, property position specifying means 2, real estate price forming factor acquisition means 3, real estate price function estimating means 4, real estate. Price function storage means 5, real estate price database 6 (hereinafter referred to as "real estate price DB"), regional information database 7 (hereinafter referred to as "regional information DB"), and real estate price function database 8 (hereinafter referred to as "real estate price function DB"). ")
It is configured to include.

The real estate price data input means 1 is data relating to the price of real estate (hereinafter referred to as "real estate price data"), that is, property address data of a property and property attribute data (construction year, It has a function to input equipment conditions such as size and presence of elevator) to this device. The input data is stored in the real estate price DB 6.

The property position specifying means 2 has a function of specifying the position (coordinates) of the property input by the real estate price data input means 1. The position of the property in this embodiment means a coordinate represented by latitude and longitude. The property position specifying means 2 uses a geographical information device (GIS), an electronic map, and a GPS (Global Positi) based on the address data of the property.
oning Systems) to identify the coordinate data of the property. The coordinate data specified by the property position specifying means 2 is stored in the real estate price DB 6, and thus the real estate price DB 6 becomes “a real estate price DB having coordinate data”.

The real estate price formation factor acquisition means 3 is based on the coordinate data specified by the property position specification means 2 and the area information database 7, and is the factor data for forming the real estate price of the property (hereinafter referred to as "factor data"). ) Is acquired or created. The factor data acquired or created by the real estate price formation factor acquisition means 3 is stored in the real estate price DB 6 having coordinate data, whereby the real estate price DB 6 displays “real estate price D having regional information”.
B ”.

The real estate price function estimating means 4 has a function of estimating a real estate price function with a real estate price DB 6 having regional information as an analysis target. Specifically, the real estate price function estimating means 4 includes a market segmenting means, a model optimizing means, and a variable optimal converting means. The market segment means has a function of identifying a segment having the same market structure by a statistical test based on the real estate price DB 6 having the regional information. The model optimizing means has a function of selecting an optimal model from a plurality of models (algorithms) for estimating the real estate price function. The variable optimum conversion means has a function of converting the variables used in the real estate price function by an optimum conversion method, a function of automatically categorizing the variables, and the like.

The real estate price function storage means 5 has a function of storing the function estimated for each segment by the real estate price function estimation means 4 in the real estate price function DB 8. That is, the values of variables and coefficients of the real estate price function are stored in the real estate price function DB8.

As described above, the real estate price DB 6 is for storing real estate price data relating to the address data of the property and the property of the property, and in the present embodiment, a newly built condominium database 6a, a used condominium database 6b, a detached house database. The housing database 6c, the rent database 6d, the land database 6e, and the other real estate price database 6f are provided. The new condominium database 6a and the land database 6e store address data of new condominiums and land and various data relating to property attributes. The other real estate price database 6f stores, for example, public land price information such as price data (actual price) based on actual transaction cases in the market, official land price / reference land price, property tax evaluation value, inheritance tax land price, and the like.

In the real estate price DB 6, after the real estate price data is input by the real estate price input means 1, the coordinate data is stored by the property position specifying means 2 and then the factor data is stored by the real estate price formation factor acquisition means 3. By doing so, it becomes the analysis data of the real estate price function estimating means 4.

The regional information DB 7 stores various data necessary for acquiring or creating factor data constituting variables of the real estate price function, and specifically, the physical region database 7a (hereinafter referred to as "physical region"). And a subjective area database 7b (hereinafter referred to as "subjective area DB"). The physical area DB 7a includes a city information database with coordinate data, a network information database, and a spatial database, and data for obtaining or generating physical factor data (for example, commercial facilities,
Data on public facilities, natural environment, land use intensity, etc.) are stored. The subjective area DB 7b stores data for acquiring or generating cultural factor data or historical factor data for each region (for example, data on expert rankings for regions, data on consumer reaction, etc.). To do.

The real estate price function DB 8 is for storing the real estate price function estimated by the real estate price function estimating means 4, and the variables and coefficient values of the real estate price function are specified for each segment specified by the market segment means. Stored in. Specifically, for example, "real estate price = a0
+ a1 x Exclusive area + a2 x Balcony + a3 x Construction + a4 x Number of rooms + a5 x Total number of units + a6 x Market residence time + a7 x Corner dummy + a8
The value of the variable and coefficient of the real estate price function represented by “commercial accumulation + a9 × accessibility + a10 × public facilities accumulation + a11 × green coverage + a12 × noise + ... + e” are stored.

Here, it is assumed that the apparatus 100 is realized by software that defines the operation of each of the above means in a processing device having a general structure including a control means, an input means, a storage means, and an output means. is doing. That is,
The program that defines the operations of the real estate price data input means, the property position specifying means, the real estate price formation factor acquisition means, the real estate price function estimating means, and the real estate price function accumulating means, is executed by the control means to Cooperate with hardware to realize various functions in the real estate price function estimation device.

The apparatus 100 can be realized by a physically dedicated apparatus, and the function of one means can be realized by two or more physical means.
It also includes the case where the functions of two or more means are realized by one physical means. Further, the device 100 may be configured by a single computer or may be configured by a plurality of computers distributed on a network.

(Processing Flow in Real Estate Price Function Estimating Device) (Real Estate Price Data Input Means) Next, the processing flow in this device will be described in detail with reference to FIG. First, the real estate price data input means 1 inputs real estate price data. FIG. 2A is an example of real estate price data input by the real estate price data input means 1. According to the figure, data such as address, condominium name, price, occupied area, and year of construction (years of construction) of the second-hand condominium property are input. The real estate price data input means 1 can input data, for example, by reading a predetermined database or storage medium storing real estate price data. The input data is registered in the real estate price DB6.

(Property position specifying means) Next, the property position specifying means 2 acquires coordinate data (latitude / longitude) of each property stored in the real estate price DB 6. FIG. 3 is a diagram for explaining the flow of the property position specifying process for specifying the coordinate data of each property. According to the figure, as a method of obtaining the coordinate data of each property, a method of using a geographic information system (GIS), an electronic map, and a GPS is described. That is, when the data of the corresponding property is read from the real estate price DB 6 (S301), the geographic information system (S302) and the electronic map (S302) based on the read address data of the property.
303) or the GPS (S304) identifies the coordinate data of the property (S305), and the identified coordinate data is stored in the real estate price DB 6 (S306).

The geographic information system (GIS) uses a database and various kinds of data input in real time as digital data.
It has a function to visually represent on the map,
It is possible to overlook the distribution and arrangement of people. In this embodiment, the coordinate data is acquired by the geocode operation of the geographical information system. Specifically, the geographical information system decodes the character address data stored in the real estate price DB 6, specifies the coordinate data corresponding to the address data, and outputs it to the real estate price DB 6.

The electronic map has a function of outputting the coordinate data of the position when a predetermined position is designated on the electronic map. In this embodiment, for example, when the person in charge of inputting the real estate price data directly plots the position of the property on the electronic map based on the address data stored in the real estate price DB, the coordinates of the property are displayed by a predetermined function of the electronic map. The data is specified and output to the real estate price DB 6.

GPS, the Global Positioning System,
It is known as a position measuring means of a car navigation system and has a function of measuring a measuring position by receiving radio waves from GPS satellites. In the present embodiment, the person in charge of inputting the real estate price data sets the real estate price DB
Go to the site where the property specified based on the address data stored in 6 exists and use the GPS locally to acquire the coordinate data of the property. The acquired coordinate data is
It is stored in the real estate price DB 6. If the GPS has a camera function or a binoculars function, the person in charge of inputting should go to a place where the property can be viewed, not at the site where the property exists. In this case, the input person is
The coordinate data can be acquired by confirming with PS or the like.

FIG. 2B is a view showing an example of the data structure of the real estate price DB 6 in which the coordinate data is stored by the property position specifying means 2. According to FIG. 3B, it can be seen that x coordinate data and y coordinate data are stored for each property.

The property position specifying means 2 acquires the nearest station data, the road distance data to the nearest station, the data of the surrounding environments, etc., which are necessary for price prediction, based on the position of the specified property. It has a function to do. In this embodiment,
The geographic information system described above is used. Specifically, when the geographic information system inputs the coordinate data of the specified property, the geographic information system searches and measures data such as the nearest station and road distance based on the coordinate data and outputs the data. The acquired data is stored in the real estate price DB 6. FIG. 2C is a diagram showing an example of the data structure of the real estate price DB 6.
According to the figure (C), it can be seen that the nearest station data acquired based on the coordinate data, the distance from the nearest station (station distance) data, and the like are stored for each property. (Real Estate Price Forming Factor Acquiring Means) Next, the real estate price forming factor acquiring means 4 acquires or creates factor data that constitutes a variable of the real estate price function. The real estate price formation factor acquisition means 4 acquires or creates factor data of each property by searching the regional information DB 7 based on the coordinate data specified by the property position specifying means 2.

FIG. 4 is a flow chart showing the flow of processing of the real estate price formation factor acquisition means 3. The real estate price formation factor acquisition means 3 acquires the coordinate data of the corresponding property from the real estate price DB having the coordinate data (S40).
1), physical area DB7 based on the acquired coordinate data
By searching the a and / or the subjective area DB 7b, area information data representing a factor that forms the real estate price is acquired (S402, S403). Then, the real estate price formation factor acquisition means 3 creates factor data for forming the real estate price by directly inputting the acquired regional information data, making a total score, or making a composite index (S4).
04). The created factor data is stored in the real estate price DB 6 (S405).

Here, the physical area DB 7a includes a town information database with coordinate data, a network information database, and a spatial database. The town information database with coordinate values associates data of commercial facilities with the coordinate data. To store. The network information database stores data such as train and bus timetables and road maps. The spatial database stores other data such as restricted areas.

The real estate price formation factor acquisition means 3 creates factor data such as commercial accumulation, public facility accumulation, natural environment, and land use intensity from the physical area DB 7a. In addition, factor data such as subjective value for the area is created from the subjective area DB 7b. The various factor data will be specifically described below.

The commercial accumulation is factor data indicating the degree of accumulation of commercial facilities such as large-scale retail stores, convenience stores, specialty stores and banks. Specifically, it is represented by the number of commercial facilities, floor area, sales, etc. within an arbitrary radius (for example, within 300 m). As the aggregation method, it is desirable to extract an optimum influence range such as within an arbitrary radius or within a radius depending on the road distance.

The public facility accumulation is factor data indicating the degree of accumulation of public facilities such as ward offices, branch offices, gymnasiums, and public halls. Specifically, the number of public facilities and the floor area within an arbitrary radius (for example, within 300 m) are applicable. As for the aggregation method, it is desirable to extract the optimum range of influence such as within an arbitrary radius or within a radius depending on the road distance, as in the case of commercial accumulation.

The factor data belonging to the natural environment corresponds to data indicating the level of the natural environment such as the green coverage, the green visibility, the air pollution degree, and the noise. Factors related to land-use intensity include land-use intensity, such as the degree of building, the type of impact of a disaster (earthquake), the probability of flood damage, and the probability of a landslide disaster.

The subjective value for a region indicates the popularity of the region, and the aggregation method can be made variable by using cultural factors or historical factors as an index that can reflect quantitative value. Specifically, a real estate sales representative,
Scores are measured through rankings by experts such as real estate agents and investors, and voting by residents and immigrants. In the present embodiment, as an example of the subjective value, a "reverberation index" that represents the response of consumers and experts to the property is set.

FIG. 8 shows an example of the real estate price DB 6 with regional information. According to the figure, for each property, commercial accumulation,
It can be seen that factor data such as accessibility (transport convenience), concentration of public facilities, green coverage, noise, reverberation index, and dummy along the railroad line (○○ Electric Railway Line) are stored.

Further, the real estate price formation factor acquisition means 3 is
The local information data is acquired and the physical factor data (hereinafter,
It is called "physical variable". ) Has the ability to create. The physical variable creating function includes, specifically, a function of estimating a variable from another data, a variable integrating function, and a function of acquiring a variable using photo data and the like.

First, the function of estimating a variable from another data is to predict the variable from data having causality, not data directly related to the variable. FIG. 5: is a figure for demonstrating the function which estimates a variable from another data. For example, variables relating to soil pollution will be described. The degree of soil pollution is an important factor in predicting real estate prices, and as a method of measuring the degree of soil pollution,
(1) There is a method of directly measuring the degree of soil pollution. However, in this embodiment, as an estimate from another data,
(2) Forecast based on the causality that there is a possibility of soil contamination if the previous land use was a factory. The variable thus predicted is set as a “soil pollution possibility dummy”.

As another example of the method of estimating the variable from the separate data, when the variable is the road traffic noise amount, in addition to directly measuring the noise amount, the road traffic amount which is the noise source and the building build-up degree There is a method such as estimating from.

Secondly, when there are a large number of variables that can be classified into similar categories, the variable integration function classifies these variables into categories and creates an overall score, and analyzes the relationship between the overall score and the real estate price. It is a thing. FIG. 6 is a diagram for explaining the variable integration function. For example, there is a close relationship between the degree of commercial agglomeration and the price of real estate, but when there are many variables that represent the degree of commercial agglomeration, the relationship with the price of real estate cannot be fully explained. There are cases. In such a case, factors that are classified into similar categories (number of convenience stores, large-scale retail stores, specialty stores, sales floor area, sales amount, etc. within a specific range) are converted into a total score, and the relationship between the total score and the price is calculated. By analyzing, the degree of commercial accumulation is calculated. The total score can be created by performing principal component analysis and factor analysis. Thirdly, the function of acquiring variables by using photographic data, etc. This is to get the variable. FIG. 7 is a diagram for explaining the function of acquiring a variable using photograph data or the like. According to the figure, first, the amount of heat etc. is analyzed from the satellite photograph data. On the other hand, greens within an arbitrary radius are measured from the navigation photo data, and greens other than trees are further removed from the measurement results. Then, the green of the natural forest is measured based on the analysis result of the heat quantity and the measurement result of the green of the natural forest. When removing greens other than trees, rivers can be removed based on national land numerical information, etc., and buildings can be removed based on GIS data with buildings.

Further, the real estate price formation factor acquisition means 3
Has an optimization function for the creation of physical variables. For example, a case where the physical variable is “distance to the nearest station (hereinafter, referred to as“ station distance ”)” of traffic convenience will be described (transport convenience).

The real estate price formation factor acquisition means 3 uses the "walking distance area", "bus area", "automobile area", etc. for "station distance",
Perform optimization so that comparisons can be made on the same scale.
Specifically, there are a method of converting a time-converted one into a distance and a method of converting into a money. An example of converting to money is shown below. Walking area: Average wage / hour x Time to the nearest station Bus area: Average wage / hour x Bus ride time to the nearest station + Bus wage Automobile area: (Average wage / hour x Car time to the nearest station) ＋ (Car cost (fuel cost, etc.) / Distance to the nearest station) Next, the physical variable is “accessibility to the city center” that is transportation convenience.
(Transport convenience) will be described. The "accessibility to the city center" can be expressed by the time distance to each main terminal station. The time distance to each main terminal station means the time distance to a representative terminal station (eg, “Tokyo station”) or the nearest terminal station (eg, “Nerima station”). Here, the real estate price formation factor creating means 3 can perform optimization by expressing the “accessibility to the city center” by a comprehensive index.

The first example of the integrated index is the time required to reach a plurality of main terminal stations. Specifically, let us consider the case where a new line has been developed for a single target station. In this case, even if there is no change in the time required to reach that station, the development of a new line may shorten the time required to reach another major terminal station, thereby increasing regional potential and increasing real estate prices. May rise. Therefore, for example, optimization can be achieved by creating a simple average of the time required to reach multiple main terminal stations in the Tokyo area and a weighted average index according to the potential such as the number of passengers at each main terminal station. .

A second example of the integrated index is an index obtained by adding the boarding charge to the time required to reach each main terminal station. That is, as the accessibility to the city center, it may be necessary to consider not only time costs but also financial costs such as boarding fees. In such a case, it becomes possible to compare them as a single index by making a monetization.
Specifically, it is calculated as ““ average wage / hour ”ד time to main terminal station ”+ boarding fee”. Further, not only the average of the above-mentioned indexes calculated for each station but also a variable for weighted average by the number of passengers can be set.

(Real Estate Price Function Estimating Means) Next, the real estate price function estimating means 4 will be described. The real estate price function estimating means 4 automatically determines a segment having the same market structure, performs a structural change test based on the determined data for each segment, and optimizes the variables and models constituting the real estate price function. By doing so, it has a function of estimating the real estate price function.

The processing of the real estate price function estimating means 4 will be described by taking as an example the case of estimating the real estate price function for predicting the real estate price in June 2001 under the following conditions. Real Estate Price Function Estimated Date = June 1, 2001 Real Estate Price DB = Data for January-April 2001 (stored) Newly input data = Data for May 2001 Estimated real estate price function = 2001 In this embodiment, the real estate price function is estimated every month based on the real estate price data updated monthly. For example, the real estate price data is updated weekly. In such cases, the real estate price function can be estimated weekly.

FIG. 8 shows the apparatus 1 as of June 1, 2001.
00 is an example of a real estate price DB 6 owned by 00. According to the figure, real estate price data for January to May 2001 are stored. All the data stored in the real estate price DB 6 (hereinafter referred to as “analysis data”) are numerical data, and continuous data such as price and occupied area and dummy variables (eg:
The corner room consists of 1, and the others are 0).

FIG. 9 is a flowchart showing the flow of processing of the real estate price function estimating means 4. First, the real estate price function estimating means 4 reads the analysis data of the real estate price DB 6 (S901), performs market segment processing (S902) to determine the optimum segment, and then uses it to create the real estate price function. A model optimization process (S903) of selecting an optimum model from a plurality of models and a variable optimization process (S904) of optimizing variables in the real estate price function are performed. Finally, based on the model selected in the model optimization process and the variable optimized in the variable optimization process, a process of estimating a real estate price function used for actual prediction is performed (S905). .

(Market Segment Processing) FIG. 10 is a flowchart showing the main flow of market segment processing.
The real estate price function estimating means 4 performs three structural change tests in the market segment processing. First, the real estate price function estimating means 4 reads the analysis data from the real estate price DB (S1011), and performs a structural change test to determine the segment based on the real estate type (S1012).
As a result, a real estate price DB with the same structure of real estate types
To confirm. Next, a structural change test is performed to determine the segment on the condition of time (S1013), whereby the real estate price DB having the same temporal structure as the real estate type is determined. Finally, the real estate price function estimating means 4 performs a structural change test to determine the segment based on the region (space) (S1014).
A real estate price DB with the same quality for each time and area structure is established. By performing the above three structural change tests, the optimum segment for estimating the real estate price function is determined (S1015).

The structural change test (stability test) in the present invention is a linear model (y = a0 + b1x1 + b2x2 ------).
-) Is to test whether the estimated parameters such as b1 and b2 are the same for the arbitrarily divided markets such as A and B. In this embodiment, AIC (Ak
aike's Information Criterion: Akaike's Information Criterion). Formula: AIC = -2 × log
AIC defined by (maximum value of likelihood) + 2 × (number of free parameters) is an estimated value of the information criterion, and the difference between the AIC values of two or more models is relative to the model. It is an index of goodness. That is, the model with the smaller AIC value is the better model. In the present embodiment, assuming that the explanatory power of the model increases as the number of analysis data increases, when the data amount increases and the AIC increases, it is considered that there is a structural change. That is, it is determined that the AIC has increased because the parameter to be estimated is not substantial.

FIG. 11 is a diagram for explaining the relationship between the segment and the real estate price function. According to the figure, first, n Ds of different real estate types from the real estate price DB 6
B is divided, and n DBs having different temporal structures with respect to the real estate type (1) are further divided, and n DBs having different regional structures with respect to the real estate type (1) time (1) are further divided.
B is divided. As a result, a segment with the same market structure is established under each condition. Then, the real estate price function is estimated by using the finally divided data for each segment as analysis data. According to the figure, the real estate price function has a one-to-one correspondence with each segment.

(Structural Change Test Conditioned on Real Estate Type) FIG. 12 is a flowchart showing the flow of processing when a structural change test is performed on the condition of real estate type. It is an object of the present invention to specify a segment based on a predetermined statistic, not to specify a segment based on an arbitrary division such as one room or family type. That is, the real estate price estimating means 4 determines the scale of the property (square meter, (hereinafter, "m2"
Say. )) Is performed, a segment with the same structure by real estate type is identified by performing a predetermined structural change test.

Specifically, first, analysis data is read from the real estate price DB 6 (S1201), and the analysis data is divided into two according to preset conditions (S120).
2) Set two data groups (S1203, S1)
205). In the present embodiment, the analysis data is converted into a data group of 10 m2 or more (DB_a) according to the condition of “scale = 10 m2”.
And a data group (DB_b) of less than 10 m2.

FIG. 13 is a diagram showing an example of the data structure of the analysis data file used in the structure change test. In the figure, the real estate price DB 6 is
It can be seen that it is divided into two data groups (DB_a, DB_b). DB_a has data only in the scale (S_01 to S_11: 1 group), and DB_b has data only in the scale (S_S to S_ (S + 1): 2 groups).

Next, the real estate price function estimating means 4 performs a regression analysis on each divided data group (S120).
4, S1206), and the result (X) is obtained (S120).
5). The real estate price function estimating means 4 also performs a regression analysis on the entire analysis data before the division (S120).
8) and the result (Y) is obtained (S1209).

Finally, the real estate price function estimating means 4 divides the result (X) and does not divide it (Y).
Is determined by AIC.

That is, the value of AIC (X) is AIC
If it is smaller than the value of (Y), it is determined that there is a structural change, and if the value of AIC (X) is not smaller than the value of AIC (Y), it is determined that there is no structural change. If it is determined that there is a structural change, the two data groups have different structures, so the division of the two data groups is confirmed (S12).
11). On the other hand, when it is judged that there is no structural change,
Since the two data groups have the same structure, the two data groups are not divided.

The real estate price function estimating means 4 determines the value at which the data should be divided and divides the analysis data into a plurality of data groups by sequentially updating the division conditions and repeating the same structural change test. For example, if the condition is set to increase by 5 m2 and it is determined that there is no structural change at scale = 10 m2, the condition is set to scale = 15 m2 and the structural change test is performed again.
If it is determined that there is a structural change at scale = 15m2, the division of the data group is confirmed at this stage, and next, 15m2
A structural change test is performed on the above data group with the condition set to scale = 20 m2.

As described above, by repeating the structural change test on the analysis data under the condition of the scale,
Real estate price DBs having the same structure in the real estate type can be specified. According to FIG. 11, it can be seen that the real estate price DB is divided into n DBs.

In the present embodiment, AIC is used as an index for determining whether or not there is a structural change in the real estate type, but the present invention is not limited to this.
IC (Bayesian Information Criterion), R-value with adjusted degrees of freedom r'2, Mallows (Malls)
It is also possible to apply an index such as Cp statistic of ows).

(Structural Change Test with Time as Condition) Next, a case where a structural change test is performed with time as a condition will be described. FIG. 14 is a flowchart showing the flow of processing when a structural change test is performed under the condition of time. In this embodiment, since the real estate price data of May 2001 is newly input, the analysis data of May 2001 is compared with the analysis data of January to April 2001 (assuming the temporal structure is the same). A case where it is judged by the structure change test whether or not the data structures are the same will be described. The structural change test with time as a condition is performed for each DB divided by the structural change test according to the real estate type.

First, the real estate price function estimating means 4 reads the analysis data (S1401), divides the analysis data according to a preset condition (S1402), and sets the data group A (DB_a) and the data group B (DB_b). And (S
1403, S1405).

Specifically, the real estate price function estimating means 4
Reads the analysis data (January to May 2001), divides the analysis data according to the condition "Time = June 2001", and
The data group A (DB_a) from January to April 1st and the data group B (DB_b) from May 2001 are set.

Next, the real estate price function estimating means 4 performs regression analysis on each data group (S1404,
(S1406), the equivalence test of the variance of the error term is performed on the result of the regression analysis of each data group (S1407). Next, the real estate price function estimating means 4 determines whether or not the variances of the error terms match as a result of the equivalence test of the variances of the error terms (S1408), and if they match, a linear constraint test is performed (S1409). ), And when they do not match, an asymptotic ratio test is performed (S1410).

Then, the real estate price function estimating means 4 performs a linear constraint test or an asymptotic ratio test, and then performs a regression coefficient correspondence test (S1411). If the regression coefficients match, there is no structural change. (S1413), 2
Since two data groups have the same structure, the two data groups are not divided. On the other hand, if the regression coefficients do not match, it is determined that there is a structural change, and since the two data groups have different structures, the division of the two data groups is confirmed. Specifically, the data group from January to April 2001 and the data group from May 2001 are divided.

(Structural Change Test Conditioned on Region) Finally, a case where a structural change test is performed on condition of the region will be described. In the present embodiment, the structure change test is performed by using “Machichome” in the administrative division as a unit of regional structure and using a region code in which a region code corresponding to each “Machichome” is set in advance. For example, Oyamadai 1-chome area code =
It is 1001, and the area code of Oyamadai 2-chome is set to 1002. The area code is stored in an area code DB (not shown). In the present embodiment, the structural change test is performed on the reference area and the area adjacent to the reference area.
Integrate two regions and do not integrate two regions if there is structural change. In the structural change test, a reference area and an area adjacent to the reference area are selected as the structural change test targets.

FIG. 15 is a flow chart showing the flow of processing in the case where a structural change test is carried out on the condition of an area. First, the real estate price function estimating means 4 reads the analysis data (S1501), and extracts the data of the area serving as the reference of the analysis target based on the area code DB (S150).
2). Next, the real estate price function estimating means 4 determines whether or not there is an area adjacent to the reference area based on the area code DB (S1503), and if the adjacent area exists, it is extracted. A structural change test is performed on an area adjacent to the area (S1504). The structural change test is performed according to the structural change test procedure described with reference to FIG.

Then, as a result of the structural change test, it is judged whether or not there is a structural change (S1505). If there is no structural change, both regions have the same structure, so both regions are integrated (S1506). . Next, the real estate price function estimating means 4 determines whether or not the analysis data is over (S150).
7) If the analysis data is not the end, it is determined whether or not there is another adjacent area with respect to the reference area, and a structural change test is performed in the same manner.

FIG. 16 is a diagram for explaining how regions are integrated as a result of the structural change test. The real estate price function estimating means 4 first uses the reference area code = 10.
When 01 is extracted, it is next determined whether or not there is an area adjacent to the area code 1001. Since the area code 1002 exists as an area adjacent to the area code 1001, the real estate price function estimating means 4 performs the structural change test on the data group of the area code 1001 and the data group of the area code 1002 by the procedure of FIG. To do.

That is, the analysis data is read (S1
401), two data groups including a data group (DB_a) of the area code 1001 and a data group (DB_b) of the area code 1002 are set (S1402, 1403, 1405), and regression analysis is performed for each (S1404, 1406). ). Then, the equivalence test of the variance of the error terms is performed on the result of the regression analysis (S1407), and if the variances of the error terms match, a linear constraint test is performed (S1409). If they do not match, the asymptotic ratio test is performed (S14).
10).

Then, the real estate price function estimating means 4 carries out a relevance coefficient correspondence test (S1411), and if the regression coefficients match (S1402), it judges that there is no structural change, and the data group A (DB_a ) And the data group B (DB_b) are combined. That is, the data group of the area code 1001 and the data group of the area code 1002 are combined. If the regression coefficients do not match, it is determined that there is structural change and the data group (DB_
Do not combine a) with the data group (DB_b).

Next, since the analysis data is not over, the real estate price function estimating means 4 judges whether or not there is another adjacent area in the area code 1001 (S1503). For example, the area code 1010 is targeted. Conduct a structural change test. Here, when the area codes 1001 and 1002 are integrated, the structural change test is performed on the areas adjacent to the area codes 1001 and 1002, respectively, using the integrated data as a reference.
If there is no adjacent area, the processing is stopped.

According to FIG. 16, it can be seen that, for the area of the area code 1001, the areas of the area codes 1002, 1011, and 1012 are combined by the structural change test to determine one segment. Also, it can be seen that the structural change test based on the area code 1001 is completed and the new structural change test based on the area code 1003 is performed when the areas of the area code 1012 are integrated. As a result, for the areas with area code 1003, area codes 1004, 1013,
And 1023 regions have been newly consolidated and another segment has been established.

As described above, in the present embodiment, since the structural change test according to the real estate type is performed according to the scale, it is possible to specify the segment that is not bound by an arbitrary definition such as one-room or family type.

Further, in this embodiment, since the structural change test with time is performed, it is possible to appropriately grasp the change of the market structure due to the change of the interest rate / tax system or the macroeconomic environment. It is possible to identify the segments having the same structure.

In addition, in the structural change test on the condition of the area, the structural change test is performed in units of towns and streets. Therefore, after detecting a change in the behavior of the supplier or the consumer in the area, a market segment with the same area structure Can be specified.

Further, in this embodiment, since the structural change is made on the condition of the real estate type, time, and area that affect the real estate price, the segment can be finely specified. It is possible to improve the prediction accuracy.

In the present embodiment, the conditions for the structural change test are set to be “real estate type, time, and region”, but the present invention is not limited to this, and is related to changes in the real estate market structure. If there is a certain factor (for example, “station distance”), it can be set arbitrarily.

(Model Optimization Process) In the model optimization process, a model used for estimating the real estate price function is selected from a plurality of models. First, a plurality of models prepared in advance are applied to the analysis data divided for each segment to create a plurality of real estate price functions, and the prediction accuracy, that is, The optimal model is selected by selecting the function with the smallest error. This makes it possible to estimate the real estate price function for which an optimal model is selected for each segment.

The model optimization process will be specifically described below. The model optimization process has a function of selecting an optimum model from a plurality of models based on the following mathematical formula 1.

(Formula 1) Real estate price = a0 + a1 x occupied area + a2 x balcony + a3 x new year + a4 x number of rooms + a5 x total number of units + a6 x market residence time + a7 x corner dummy + a8 x commercial accumulation +
a9 x accessibility + a10 x public facility accumulation + a11 x green coverage + a12 x noise + ... + e
Least squares method, maximum likelihood method, neural network, covariance structure analysis, GMM, Spatial Statistics, etc.
These models can be set arbitrarily.

FIG. 17 is a flowchart showing the flow of model optimization processing. The real estate price function estimating means 4 is
First, the test price function i is constructed, then the test price function is verified, and finally the prediction price function is estimated.

That is, first, the real estate price function estimating means 4 constructs the test price function i (S170).
1). Specifically, the test analysis data is read from the real estate price DB 6 (S1702), and a plurality of models (i =
The test price function i is estimated for n) (S170).
3).

Next, the real estate price function estimating means 4 verifies the test price function i (S1704). Specifically, the real estate price function estimating means 4 uses the real estate price DB.
The verification analysis data is read from S6 (S1705), the deviation rate of the test price function i is calculated, and then the fitting rate is calculated based on the calculated deviation rate (S1706).

The deviation rate is the actual transaction price = MP in the verification analysis data and the estimated price by the test price function i = PPi (i = 1 ...
... n) is calculated and calculated by the following formula.

Deviation rate = (MP / PPi) × 100 If the value of the deviation rate is 100, the transaction price and the estimated price by the model match. Therefore, the closer the deviation rate value is to 100, the more accurate the prediction accuracy is. Will be high.

The Fitting Rate, which indicates the matching rate between the predicted value and the actual sales price, is calculated by the following equation and
The closer the value of ing Rate is to 0, the higher the prediction accuracy.

Fitting Rate = (average deviation rate-100)
Therefore, the real estate price estimation means 4 selects the model whose Fitting Rate value is closest to 0 as the optimum model (S1707).

Finally, the real estate price function estimating means 4 carries out a forecast price function estimating process (S1708). Specifically, the forecast analysis data is read from the real estate price DB 6 (S1709), and the forecast price function is estimated by the method of the model selected in S1707 (S1710).

The model optimizing process in estimating the price function for calculating the real estate price in June 2001 will be specifically described below.

First, the real estate price function estimating means 4 reads the data of "January 2001 to April 2001" from the real estate DB 6 as test analysis data, and a plurality of models (i = 1 ...
n) is performed to estimate the test price function i. In this embodiment, i = 26, that is, 26 models are prepared.

Next, the real estate price function estimating means 4 reads the actual transaction price data of "May 2001" as the analysis data for verification, and uses the test price function i from the data of "May 2001". By calculating the estimated price, the deviation rate of the test price function and the fitting rate are calculated. The analysis data for "May 2001" is assumed to consist of 2630 transactions.

FIG. 18 is a diagram showing an example of the distribution of the deviation rate of the test price function i with respect to the analysis data of May 2001. According to the figure, the value of the model 15 is closest to 100, and it is understood that the model 15 is an optimal model with high prediction accuracy.

FIG. 19 is a diagram showing an example of the distribution of the Fitting Rate of the test price function i with respect to the analysis data of May 2001. According to the figure, Fittin of model 15
It can be seen that model 15 is the optimum model because g Rate is closest to 0.

Finally, the real estate price function estimating means 4 adopts the model 15 to obtain "2001
The forecast real estate price function for June 2001 is estimated for the data from January to May.

(Variable Optimal Conversion Process) The variable optimal conversion process is a process executed in the model optimization process described above. The function of performing optimal variable conversion for various variables and the categorization of dummy variables are automatically performed. It has a function to perform the target.

The function for performing the optimum variable conversion is, specifically, for continuous data, logarithmic conversion, exponential conversion, and BO conversion.
Data conversion is performed based on multiple conversion methods such as X-COX conversion, and the conversion method with the highest prediction accuracy is automatically adopted from among the multiple conversion methods. In each conversion method, only appropriate variables that affect the real estate price are extracted.

The function of automatically categorizing dummy variables is to extract an optimum grid when creating dummy variables (“corner dummy”, “railway dummy”, etc.) based on categorical data. , To set the appropriate variables. For example, for the "walking dummy",
When the distance from the location of the property to the station is less than the specified time, "1" is set, and when it is set to a value other than "0", the optimum specified time (for example, "10 minutes") is automatically set. It is something to set. As a result, even if it is an image-like variable such as “walking area” and it is difficult to define it, the image is specified by searching for the specific condition of “walking area”. To create the appropriate variables. In addition, when there is an image of "Southward is good", "Southward dummy" is created as a dummy variable, "1" is set if the property is facing south, and "0" is set otherwise. It In such a case, there are several patterns that can be considered as facing south ("South", "South" +
"Southeast", "South" + "North-South", or "South" + "Southeast"
Appropriate variables can be created by searching for the optimum pattern from (+ North / South, etc.). If the optimum pattern is not searched, the dummy variable is not set.

As described above, various estimation methods are further applied to the analysis objects whose structures are considered to be the same in the linear structure based on the constant statistics, and the variables are also optimized in the model optimization processing. Because,
It is possible to select the model with the highest prediction accuracy.

(Real Estate Price Function Storage Means) The real estate price function storage means 5 stores real estate price function data estimated for each market, that is, variable data and coefficient data in the real estate price function DB 8. The real estate price function is updated at the timing when the real estate price data is updated, and stored in the real estate price function DB 8. Thereby, for example, when the real estate price prediction device (not shown) receives a price prediction request for a predetermined property, the real estate price function of the specified segment is identified from the real estate price function DB 8 by identifying the segment to which the property belongs. By reading it, it becomes possible to use it to predict the price of the property. In addition,
This usage method will be described in the second embodiment.

The above embodiments are examples for explaining the present invention, and are not intended to limit the present invention only to the above embodiments. The present invention can be implemented in various forms without departing from the gist thereof. For example, in the above embodiment, the structure change test in the market segment processing is performed on the condition of the real estate type, time, and region, but the present invention is not limited to this, and may affect the real estate price. The conditions can be set arbitrarily. <Second Embodiment> The second embodiment of the present invention will be described below.
A description will be given with reference to the drawings. The second embodiment will explain a real estate price predicting device that predicts the price of a desired real estate property (hereinafter referred to as “evaluation target property”) by using the real estate price function estimating device according to the first embodiment. .

FIG. 20 is a block diagram showing the functional arrangement of the real estate price forecasting apparatus according to the second embodiment. The real estate price prediction device 200 is a property data input means 2 for evaluation.
01, the evaluation target property position specifying means 202, the evaluation target property market segment means 203, the living environment assessment means 204, the real estate price predicting means 205, and the prediction result output means 206, which are accessible to the real estate price function estimating device 100. It is configured.

The evaluation target property data input means 201 inputs data relating to the evaluation target property, specifically, address data and property attribute data relating to the evaluation target property. The evaluation target property position specifying means 202 specifies the coordinate data of the input evaluation target property. As described in the first embodiment, the coordinate data is specified by the geographical information system,
Electronic maps and GPS can be used.

Market segment means 203 of the property to be evaluated
Specifies the segment to which the evaluation target property belongs based on the input property attribute data of the evaluation target property and the specified coordinate data, and the real estate price function estimation device 100
Load the real estate price function corresponding to the segment identified from.

The living environment assessment means 204 acquires and indexes living environment data of the property to be evaluated based on the specified coordinate data and the area information database 7,
Set the variables used in the real estate price function. Regarding acquisition and indexing of living environment data and setting of variables,
The function of the real estate price formation factor acquisition means 3 described in the above embodiment can be used.

The real estate price predicting means 205 is a market segment means 203 of the evaluation target property for the evaluation target property.
The price of the property to be evaluated is predicted by executing the real estate price function specified by. The prediction result output unit 206 outputs the result of the price prediction of the property to be evaluated, and also outputs the data regarding the basis of the price prediction.

Next, the operation in the case of using the real estate price predicting apparatus 200 to predict the property to be evaluated will be specifically described. In the second embodiment, a case will be described in which a real estate price prediction device 200 is used to predict an evaluation target property when a financial institution lends a predetermined property (used condominium).

FIG. 21 is a diagram showing an example of a real estate price data input screen. According to the figure, the input screen is provided with items 21a such as address, applicant, real estate, loan determination, and reception, so that the person in charge can input data for each item related to the property to be evaluated. It is configured.

When the user inputs the data of the property to be evaluated and selects the "market price calculation" button 21b, the real estate price predicting apparatus 200 executes the above-mentioned means to make the real estate of the property to be evaluated. Estimate the price.

FIG. 22 is a diagram showing an example of the real estate price prediction result table output by the prediction result output means 15. According to the figure, a peripheral diagram of the evaluation target property, conditions regarding the evaluation target property, and a list of peripheral cases of the evaluation target property are output. Various conditions input by the user on the real estate price data input screen are output in the condition column related to the property to be evaluated, and the market unit price calculated by this device and the market unit price The market price is output.

Looking at the real estate price data input screen, "market unit price = 23.34 (10,000)" and "market price = 1,400"
(10,000) ”is entered. On the other hand, in the real estate price forecast result table screen, “computed market unit price = 24.58 (10,000)” and “calculated market price = 1,481 (10,000)” are output.
As a result, the person in charge knows that the value of the property is calculated to be almost the same as the value currently judged by the user or slightly higher than the value currently judged by the person in charge. Moreover, the person in charge can confirm the list of surrounding cases together with the surrounding map, so that it is easy to confirm the property to be evaluated.

FIG. 23 is a diagram showing an example of the prediction data output by the prediction result output means 15 in association with the real estate price prediction result table. The figure shows trends in the price of used condominiums in the forecast area. Specifically, it shows the price trends of used condominiums in the three regions from September 1989 to March 2001. The person in charge can read that the price of used condominiums has risen sharply as of September 1989 when the property to be evaluated was built, but the price has declined as of March 2001.

FIG. 24 is a diagram showing an example of the prediction basis material output by the prediction result output means 15. According to the figure,
It is possible to understand which variable (real estate price formation factor) affected the price of the property to be evaluated. Also,
Since the scores of the comparison target properties are shown at the same time, it is easy to understand the characteristics of the evaluation target properties. For example, the evaluation target property has higher scores for “accessibility to the city center” and “educational environment” than the comparison target property, but is lower for other factors.

FIG. 25 is a diagram showing another example of the prediction basis data output by the prediction result output means 15. In the figure, the relationship between the age of the apartment and the price is graphed. The person in charge can review the forecast result based on the type of segment to which the property to be evaluated falls and the current age of the building.

Although the evaluation target property position specifying means 11 has been described as determining the evaluation target property by the coordinate data, the present invention is not limited to this and can be determined by another method. Specifically, the nearest station,
Conditions such as the distance to the nearest station, the proximity to the city center, and the environmental value to be emphasized can also be confirmed by inputting them as characters or images.

[0131] For example, "proximity to downtown = 50 to Shibuya
Minutes "" Environmental value to prioritize = park exists within 10 minutes on foot "
If such a condition is input, the market segment means 12 for the property to be evaluated specifies a segment and a real estate price function that meet both conditions by searching the real estate price DB 6 for a property that meets the above conditions.

According to the above, the person in charge can obtain the estimated price of the property just by inputting the data of the desired property, and the basis for the price prediction is presented, so that the calculated price is It becomes easy to judge whether it is proper or not.

<Third Embodiment> A third embodiment of the present invention will be described below with reference to the drawings. The third embodiment will explain a real estate management system that predicts and manages the price of a real estate (property) using the real estate price function estimated in the first embodiment.

(Functional configuration diagram of real estate management system) FIG.
FIG. 6 is a block diagram showing the configuration of a real estate management system according to the third embodiment of the present invention. The real estate management system shown in the figure includes a real estate price prediction device 300, a user 400 (a refinancing request user 400a, a sale request user 40).
0b, purchase request user 400c), and real estate company 500
And a bank 600. Real estate price prediction device 300
Is the real estate price function estimation device 100 and the property management means 30.
1, the echo management means 302, the property database 303, the echo database 304. The real estate price function estimation device 100 is the same as the device shown in FIG. 1 above. Real estate price prediction device 300, client of the system (user 400 (400a to 400c), real estate company 5)
00 and the bank 600) send and receive data by various communication means such as a network communication system and mail.

When the real estate price forecasting apparatus 300 receives a request for price forecast and sale of real estate property from each user, the real estate price forecasting device 300 predicts the price of the property to be evaluated, saves the result together with related information, and stores the result. It has a function to notify. When the real estate price prediction device 300 receives the intermediation request from the user desiring to sell 400b, the real estate price prediction device 300 predicts the price of the property to be evaluated and requests the real estate company 500 for the intermediary request. In this way, the user 400 who wants to sell is
b and the real estate company 500 are configured to be communicable via the real estate price prediction device 300. Further, the real estate price prediction device 300 notifies the real estate company 500 of the application information when receiving the application for the property from the purchase requesting user 400c. In this case, the real estate price prediction device 300
When accepting a loan application from the purchase requesting user 400c or the real estate company 500, can transmit the price prediction result and examination data of the property to the bank 600.
In this way, the purchase requesting user 400c and the real estate company 50
0 and the bank 600 are communicable via the real estate price prediction device 300. .

The property management unit 301 executes processing according to the access contents (inquiry, request for material) by the client. When the property management unit 301 receives the price prediction processing of the property to be evaluated, the property price function estimation device 10
0 is started, and the price prediction process of the evaluation target property is executed.
The real estate price function estimating device 100 uses the property information output from the client, such as the property address, size, building age,
The price prediction process is executed with reference to information such as the number of floors. The property management unit 301 stores the result of the price prediction process in the property database 303. When the real estate price function estimating device 100 provides the real estate price DB 6 of the real estate price function estimating device 100 with the property information for identifying the property to be evaluated, the client only needs to input the address as the property information. When the property management unit 301 receives the process of browsing and purchasing the property, the property database 30
3 is accessed, the property that matches the search condition specified by the client is specified, and the result is notified to the client. The property management unit 301 can further execute a loan application process for the bank 600 as the purchase process is executed.

The property database 303 is provided with property information relating to real estate properties such as condominiums and detached houses that are desired to be sold, and land. Furthermore, the property database 303 is
Real estate company 5 accessing real estate price forecasting device 300
00 and bank 600 and their service content.

The echo management means 302 is used by the purchase requesting user 4
The access history of 00c is managed. Echo management means 30
2 stores information in the echo database 304 in association with the information specifying the purchase requesting user 400c as the access source and the access content. Based on the information stored in the echo database 304, the echo management means 302 includes, for example, inquiries about the properties registered in the property database 303 and statistics such as the degree of market echo due to the number of requests for materials, the frequency of requests for sale and purchase, and the like. Information can be collected.

When the user 400 requests a loan application from the user 400, the property management unit 301 notifies the bank 600 of the property information stored in the property database 303 and the estimated price of the property together with the input personal information.
The information notified to the bank 600 is, for example, reference information when predicting the price of a property (related transaction case (actual price), public land price / standard land price, property tax evaluation amount, inheritance tax land price) , Trend information (related price trend cases, market information, etc.) and the like.

When a customer who wishes to make a loan with real estate as collateral arrives at the bank 600, the bank 600 can access the real estate price prediction device 300 to predict the price of the collateral property. For example, in the information system of the bank 600, the real estate price prediction device 300
Is provided so as to be communicable, and a function of controlling whether or not to make a loan and determining a loan amount, that is, an examination function is provided. In the bank 600, when the property information specifying the collateral property is input by this examination function, the property information is notified to the real estate price prediction device 300. The property management unit 301 of the real estate price prediction device 300 executes price prediction processing based on the notified property information.

The property management means 301 notifies the bank 600 of the result of the price prediction process. Property management means 301
With respect to the price prediction process requested by the bank 600, since the property information cannot be viewed outside the bank, the process of storing the property information in the property database 303 can be omitted.
The bank 600 that has received the price prediction result can calculate the loan amount by referring to the price prediction result by controlling the examination function. When calculating the loan amount, the examination function uses the loan amount determination rule information (financing condition) set in the bank 600, for example, the transaction status information (deposit balance information, income information, asset information, etc.) of the customer, the loan limit amount. Refer to information (80% upper limit of price forecast price, etc.) and option information (limit adjustment by interest rate adjustment, limit adjustment by fee adjustment, etc.). These pieces of information are registered on the information system as reference information and customer information at the time of financing.

The examination function provided in the bank 600 may have a function of executing a washing process, for example. Here, the washing process means a process of re-evaluating the collateral value of the property held by the bank 600. The examination function reads property information about a certain mortgage property from the mortgage database provided in the bank 600 and notifies the real estate price prediction device 300 of the property as a renewal process. In the real estate price prediction device 300, the price prediction process is executed, and the price prediction result is stored in the bank 60.
0 is notified. The examination function refers to the notified price prediction result and stores a new collateral value for the collateral property, that is, the price prediction amount in the collateral database. At this time, in addition to the overwrite storage, the change history can be stored as difference data. In addition, the examination function may have a function of notifying the examiner when a difference of a predetermined value or more occurs between the new price estimate and the latest price estimate stored in the collateral database. it can.

(Workflow of Real Estate Price Prediction Device) Next, a workflow using the real estate price prediction device 300 will be described. In the present embodiment, the sale requesting user 40
0b will explain the workflow when the real estate is sold.

First, the user 400b desiring to sell inputs the property information (address, area, etc.) related to the property to be evaluated from a predetermined screen on the system to evaluate the property owned by him / her and the type of handling (eg, intermediary). request)
To the real estate price prediction device 300 (26
a). Property management means 301 of the real estate price prediction device 300
Accepts the property evaluation and the intermediary request from the user desiring to sell, and requests the intermediary request from the real estate company 500 (26
b). The property management unit 301 can execute the price prediction process and store the execution result in the property database 303 with reference to the input property information. The property management unit 301 can present the user with the list of real estate companies stored in the property database 303 and the service contents thereof to prompt the user to specify the handling type and select the real estate company.

On the other hand, if the user desiring to sell 400b wants to intervene in the desired real estate company, the real estate company that accepts the intermediary inputs the information on the property to be evaluated from a predetermined screen on the Web, It is also possible to request the real estate price prediction device 300 to evaluate the property (26c).

Upon receiving the evaluation of the property from the real estate company 500, the real estate price predicting device 300 activates the real estate price function estimating device 100 based on the input property information. When the price prediction process is completed, the real estate management means 301 sends the price prediction result to the real estate company 500 (2
6d), and the price forecast result in the property database 303
To store. The property management unit 301 can transmit the price prediction result to the real estate company 500 such as proximity examples (related transaction examples) and trend information (related price trend examples, market information, etc.) that are reference information.
(26d).

Upon receiving the reference information, the real estate company 500 considers the contents of the reference information and then the sale-requesting user 400
You can enter into an intermediary contract with b. Then, a price forecast material and a sales plan for the property to be evaluated are created and transmitted to the selling user (26e).

When the property management unit 301 has an advertisement processing function, the real estate company 500 can give an instruction to post on the Web or a magazine in order to advertise the property to be evaluated. (26f). In this case, the property database 30
3 can store, for example, advertisement submission information that specifies properties to be advertised. Property management means 301
Upon receiving the advertisement placement instruction, the property database 303
To extract the property information to be advertised, execute posting on the Web managed by the real estate price prediction device 300 or notification of submission to the submission window, and set advertisement on the Web or magazine. . In addition, the purchase requesting user 400c refers to the advertisement and indicates the purchase request by the real estate price prediction device 3
When 00 is notified, the content of the purchase request, for example, the property specifying information and the personal information are stored in the echo database 304. The echo management unit 302 includes the echo database 30.
When the personal information stored in 4 includes an e-mail address, for example, the user 40 who wishes to purchase the content of the advertisement.
An email can be sent to 0c. The echo management unit 302 can have a search function of searching the property database 303 based on the desired purchase conditions notified from the desired purchase user 400c, for example. In this case, the echo management unit 302 executes the search function to
It is possible to specify the sale property that matches the purchase request condition of the purchase request user 400c.

By the way, the echo management means 302 can refer to the echo database 304 and transmit the echo data to the real estate company 500 (26g). The real estate company 500 can refer to the received echo data to create a sales status report for reporting the sales status, and send the sales status report to the user desiring to sell (26h).

On the other hand, the purchase requesting user 400c who wishes to purchase the real estate can request the real estate price prediction device 300 to apply for the real estate by designating the desired real estate from a predetermined screen on the Web ( 26i). When the real estate price prediction device 300 receives an application for a property from the purchase requesting user 400c via the Web, the real estate price prediction device 300 stores personal information and purchase desired conditions regarding the purchase requesting user 400c in the echo database 304, and also applies to the real estate company 500 for the property. It can be sent that there was. 40 users who want to purchase real estate
0c can also directly apply for a property to the real estate company 500 by referring to the echo database 304 (26j).

The real estate company 500 that has received the application for the property via the real estate price predicting apparatus 300 or directly from the user 400c who wants to purchase, the loan examination if the instruction of loan examination is included in the desired purchase condition of the user 400c who wants to purchase. Is applied to the bank 600 (26k). The property management means 301 of the real estate price prediction device 300
However, if it has a loan examination application function, the purchase requesting user 400c can apply for a loan examination to the real estate price prediction device 300 (29l). In this case, the property management unit 301 obtains property information and price prediction results regarding properties targeted for application from the property database 303.
Then, the personal information of the purchase requesting user 400c and the purchase request conditions are read from the echo database 304 and transmitted to the bank 600. The bank 600 that has received the application for the loan examination executes the processing of the loan examination based on the control of the examination function.

Bank 600 accepting application for loan examination
Performs an examination based on the received information and transmits the examination result to the real estate company 500 (26m). According to the examination result, the real estate company 500 concludes a sales contract for the property with the purchase requesting user 400c. After that, the purchase-requesting user sends the necessary documents for the loan to the bank 600 by mail,
Make a loan contract with a bank (26n). The bank 600 that has made the loan contract stores information about the property that is the subject of the loan in the collateral database held by the bank. In addition,
When the examination request is received from the real estate price prediction device 300, the examination function sends the examination result to the real estate price prediction device 30.
Send to 0. When the property management means 301 of the real estate price prediction device 300 receives the screening result, the property purchasing means 400c who should notify the screening result based on the property information included in the screening result and the information of the purchasing requesting user 400c.
Is specified and the examination result is notified by e-mail, for example. By the way, the bank 600 transmits the information of the property stored in the collateral database to the real estate price prediction device 300 every year in order to wash the property (26o). When the real estate price prediction device 300 receives the property information, the real estate price prediction device 300 performs collateral evaluation of the property and transmits the collateral value at the present time (26p).

According to the above system configuration and processing, the bank 600 can access the real estate price prediction device 300 and quickly calculate an accurate collateral value. Therefore, the loan processing and the washing processing can be executed promptly with reference to the collateral value. In addition, the real estate company 500
Through the real estate price prediction device 300, the property to be sold and the user desiring to purchase can be quickly grasped, and further, the response according to the echo, for example, the price reduction process when the echo is small can be promptly executed. Further, since the loan examination procedure can be performed by referring to the property information and personal information, it is not necessary to input the information necessary for the loan examination each time.

[0154] The real estate company 500 only gives an advertisement placement instruction to the property entered after the intermediary contract, and the We
The advertisement of the property can be posted on b or paper media.

[0155]

According to the present invention, a structural change test is performed on data relating to the price of real estate properties, segments with the same market structure are set, and a model or variable of each set segment is set. Since the choice is made, it is possible to set the optimal segment that adapts to the real estate market structure, and it is also possible to select the model and variables that reflect the real estate market structure. Will be able to estimate.

[Brief description of drawings]

FIG. 1 is a block diagram showing a functional configuration of a real estate price function estimation device according to the present embodiment.

FIG. 2 is an example of real estate price data input by real estate price data input means 1.

FIG. 3 is a diagram for explaining a flow of a property position specifying process for specifying coordinate data of each property.

FIG. 4 is a flowchart showing a processing flow of the real estate price formation factor acquisition means 3.

FIG. 5 is a diagram for explaining a function of estimating a variable from another data.

FIG. 6 is a diagram for explaining an integrated function of variables.

FIG. 7 is a diagram for explaining a function of acquiring a variable using photograph data or the like.

FIG. 8 is an example of a real estate price DB 6 that the device 100 has as of June 1, 2001.

FIG. 9 is a flowchart showing a processing flow of the real estate price function estimating means 4.

FIG. 10 is a flowchart showing a main flow of market segment processing.

FIG. 11 is a diagram for explaining a relationship between a market segment and a real estate price function.

FIG. 12 is a flowchart showing the flow of processing when a structural change test is performed on the condition of real estate type.

FIG. 13 is a diagram showing an example of a data structure of an analysis data file used in a structure change test.

FIG. 14 is a flowchart showing the flow of processing when a structural change test is performed under the condition of time.

FIG. 15 is a flowchart showing the flow of processing when a structural change test is performed on the condition of an area.

FIG. 16 is a diagram for explaining how regions are integrated as a result of a structural change test.

FIG. 17 is a flowchart showing the flow of model optimization processing.

FIG. 18 is a diagram showing an example of the distribution of the deviation rate in the analysis data of May 2001.

[Figure 19] Fitting R of the price function for testing in May 2001
It is a figure showing an example of distribution of ate.

FIG. 20 is a block diagram showing a functional configuration of a real estate price prediction device according to a second embodiment.

FIG. 21 is a diagram showing an example of a real estate price data input screen.

FIG. 22 is a diagram showing an example of a real estate price prediction result table output by the prediction result output means 15.

FIG. 23 is a diagram showing an example of a prediction material that the prediction result output means 15 outputs together with the real estate price prediction result table.

FIG. 24 is a diagram showing an example of the prediction basis material output by the prediction result output means 15.

FIG. 25 is a diagram illustrating another example of the prediction basis material output by the prediction result output unit 15.

FIG. 26 is a block diagram showing a functional configuration of a real estate price prediction system according to a third embodiment. 1 ... Real estate price data input means 2 ... Property position specifying means 3 ... Real estate price formation factor acquisition means 4 ... Real estate price function estimation means 5 ... Real estate price function storage means 6 ... Real estate price DB 7 ... Regional information DB 8 ... Real estate price function DB 100 ... Real estate price function estimation device

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Takeshi Tamura             8-4-17 Ginza Stock Market, Chuo-ku, Tokyo             Company recruitment (72) Inventor Hitoshi Seike             8-4-17 Ginza Stock Market, Chuo-ku, Tokyo             Company recruitment (72) Inventor Taiji Asami             1-27-16-A Sangenjaya, Setagaya-ku, Tokyo (72) Inventor Kiyohiko Nishimura             1-2-26-403 Higashigotanda, Shinagawa-ku, Tokyo

Claims (11)

[Claims] [Claim 1] In a price function estimation method for calculating a predicted price of a real estate property by statistical analysis, a structural change test is performed on the data related to the price of the real estate property to determine the price of the real estate property with the same market structure. Setting a segment consisting of data, selecting a statistical model used for estimating the price function from a plurality of statistical models for each of the set segments, and setting the real estate for each of the set segments. Estimating the price function using the selected statistical model with respect to the data related to the price of the property, and estimating the price function. 2. The step of setting the segment comprises performing a structural change test on the data relating to the price of the real estate property under the condition of the real estate type, and thus the first data consisting of price data having the same structure for the real estate type. By setting a segment and performing a structural change test on the data regarding the price of the real estate property belonging to the first segment under the condition of time, the data regarding the price of the real estate property with the same structure with respect to real estate type and time By setting a second segment consisting of, and performing a structural change test on the data related to the price of the real estate property belonging to the second segment by region, the structure of real estate type, time, and region is Setting a third segment consisting of data on the prices of real estate properties of the same quality And estimating the price function, the step of estimating the price function is to estimate the price function using the selected statistical model with respect to the data regarding the price of the real estate property belonging to the third segment. The price function estimating method according to claim 1. 3. The price function estimating method according to claim 2, wherein the real estate type is a scale of real estate. 4. The step of selecting the statistical model includes the steps of creating n test price functions by using n statistical models for the data on the price of the real estate property, and the data on the price of the real estate property. The divergence rate is calculated based on the transaction price of the real estate property in and the estimated price estimated by the created n price functions for testing, and the prediction of the created price function for testing is calculated from the calculated divergence rate. A step of calculating a Fitting Rate indicating accuracy, and a step of selecting a statistical model having the highest prediction accuracy from the n statistical models based on the calculated Fitting Rate. The price function estimating method according to claim 1. 5. A step of selecting an optimum variable conversion method used for estimating the price function from a plurality of variable conversion methods for each of the set segments, and the real estate property for each of the set segments. Estimating the price function using the selected statistical model and the selected variable transformation method for the data relating to the price of the variable according to any one of claims 1 to 4. Price function estimation method. 6. A step of inputting address data of a real estate property and property attribute data indicating an attribute of the property, a step of specifying coordinate data of the real estate property based on the input address data, and the specified coordinates. Further comprising the step of creating factor data that forms the price of the real estate property based on data and a database that stores regional information, wherein the data relating to the price of the real estate property includes the address data and the property attribute. 6. The price function estimating method according to claim 1, further comprising data, the specified coordinate data, and the created factor data. 7. The method further comprising the step of inputting data on the price of the real estate property, wherein the price function is estimated according to the frequency of inputting the data on the price of the real estate property. The method for estimating the price function according to any one of 1 to 6. 8. A price function estimating device for calculating a predicted price of a real estate property by statistical analysis, by performing a structural change test on the data relating to the price of the real estate property, Means for setting a segment consisting of data, means for selecting a statistical model used for estimating the price function from a plurality of statistical models for each of the set segments, and the real estate for each of the set segments Means for estimating the price function using the selected statistical model for data relating to the price of the property, and a price function estimating apparatus. 9. A price function estimation program for calculating a predicted price of a real estate property by statistical analysis, wherein a structural change test is performed on the data related to the price of the real estate property, so that A function of setting a segment consisting of price-related data, for each of the set segments, a function of selecting a statistical model used for estimating the price function from a plurality of statistical models, and for each of the set segments, A program for estimating a price function, characterized by causing a computer to realize a function of estimating the price function using the selected statistical model with respect to data relating to the price of the real estate property. 10. A structure change test is performed on data relating to the price of a real estate property to set a segment consisting of data relating to the price of a real estate property having the same market structure, and a plurality of statistics are set for each of the set segments. A statistical model used for estimating the price function is selected from the models, and the price function is estimated using the selected statistical model for the data regarding the price of the real estate property for each of the set segments, Storage means for storing the estimated price function, input means for inputting address data and property attribute data of the evaluation target property, and means for specifying coordinate data of the evaluation target property based on the input address data Based on the input property attribute data and the specified coordinate data, the section to which the evaluation target property belongs Based on the database that stores the specified coordinate data and the area information, the index of the living environment of the property to be evaluated. Means for calculating as, the input property attribute data, the calculated index,
And a means for predicting a predicted price of the property to be evaluated using the read price function, a real estate price predicting apparatus. 11. A price function estimation program for calculating a predicted price of a real estate property by statistical analysis, wherein a structural change test is performed on the data related to the price of the real estate property, so that A function of setting a segment consisting of price-related data, for each of the set segments, a function of selecting a statistical model used for estimation of the price function from a plurality of statistical models, and for each of the set segments, And a function for estimating the price function using the selected statistical model with respect to the data related to the price of the real estate property, the computer having the function for estimating the price function. A method of inputting property information that specifies the property for sale, the input property information, and the estimation function of the price function. With the price function calculated by performing a gram,
A means for calculating the price of the collateral property or the sold property, and means for calculating the loan amount or the sale amount of the collateral property or the sold property from the calculated price and the loan condition or the sale condition. A characteristic real estate management system.