JP7071948B2 - Programs, equipment and methods for estimating the occupancy rate of real estate in the target area - Google Patents

Description

The present invention relates to a technique for estimating the occupancy rate of real estate in a target area.

The occupancy rate of real estate is extremely important as an index of property value. For investors and related parties (banks, rating agencies, securities companies, etc.) such as real estate investment funds, it is an index for investment decisions. As a real estate investment fund, there is a J-REIT (Japan Real Estate Investment Trust). This is a securitized financial product that substitutes for stocks and bonds with multiple commercial real estate as backing assets. It is listed on the stock exchange and is bought and sold every day. As important information for REIT evaluation, there is "occupancy rate" as "property value".

For example, in each real estate such as a hotel, the number of staying users changes every day. Here, the "occupancy rate of real estate" is considered to have a correlation with the "number of staying users" in the target area of the real estate. It is assumed that a large number of staying users means that the occupancy rate of real estate is high, and a small number of staying users means that the occupancy rate of real estate is also low.

There is a technique for counting the number of users in the area by collecting the number of staying users, for example, the position of a mobile terminal owned by the user (see, for example, Patent Document 1). In general, the subscriber ratio of a particular carrier to all carriers is used to multiply the number of users connected to the carrier's communications equipment for a relatively large area by the subscriber ratio. By doing so, it is possible to estimate the approximate number of staying users (for example, in units of tens of thousands). This is commonly referred to as the "expanded estimation method". For example, when estimating the population movement of tens of thousands of people in the event of a disaster, if the position movement of 1% of users can be captured, the position movement of hundreds of people can be estimated.

WO2012 / 036222

Here, a sufficient population is required to grasp the change in the number of staying users in each region. In particular, the smaller the area such as real estate, the smaller the number of captured users and the larger the variation in the estimated number of staying users.

FIG. 1 is an explanatory diagram showing the number of captured users in a specific telecommunications carrier for each region.

According to FIG. 1, the number of captured users connected to the telecommunications business equipment of a specific telecommunications carrier is shown for each real estate. The number of captured users by a particular carrier can only be counted to a much smaller number than the actual number of staying users in the area. That is, it is not possible to collect the location information of all users from all telecommunications carriers.

For example, in a small area of real estate such as a hotel, it is not possible to simply use the subscriber ratio for each carrier in order to estimate the number of staying users.
For example, when estimating the number of staying users for a hotel with an allowable number of rooms of 100, even if the actual number of staying users is 10, the number of captured users in a specific telecommunications carrier may be 0.
If the subscriber ratio of a specific telecommunications carrier is, for example, 20%, if the number of captured users is 0, the number of staying users at the hotel is estimated to be 0, and if the number of captured users is 1. The number of users staying at the hotel is estimated to be five.
Similarly, for a particular carrier, even if the number of captured users at the hotel is 3, it does not mean that the number of staying users is estimated to be 15 according to the subscriber ratio.
As described above, in real estate in a relatively small area, the difference in the number of captured users greatly affects the estimation of the number of staying users. With such an unstable number of captured users, it is not possible to estimate the number of staying users.

Considering this, it is very difficult to estimate the number of staying users in real estate in a small area by the conventional expansion estimation method.
On the other hand, for real estate investors and related parties, the only way to accurately know the number of staying users, which leads to the occupancy rate of real estate, is to wait for the announcement of the number of users for each fiscal year. On the other hand, there is a risk of fluctuations in the occupancy rate of real estate, and this opacity also increases the difficulty of investment based on the property value.

Therefore, the present invention presents a program, an apparatus, and a device capable of estimating the occupancy rate of real estate in a relatively narrow target area as accurately as possible even if the number of captured users is connected to the communication equipment of a specific telecommunications carrier. The purpose is to provide a method.

According to the present invention, it is a program for operating a computer mounted on a device for estimating the occupancy rate Ci of real estate in the target area Ai.
A user position database that accumulates time-series positions for each user who has a mobile terminal that can acquire the position,
A regional characteristic quantity storage means for storing one or more characteristic quantities of the region A for each region A,
Using the user location database, the number of captured users #A that stays in each region A for a predetermined period T is extracted, and the number of captured users is extracted.
Using the regional characteristic quantity storage means, each weighting coefficient ω according to the degree of similarity d between the characteristic quantity of the target area Ai and the characteristic quantity of a plurality of N surrounding areas Aj (j = 1 to N) with respect to the target area Ai. A weighting coefficient determining means for determining (Ai, Aj) and
Peripheral area capture calculated by weighting the weighting coefficient ω (Ai, Aj) to the number of captured users #Aj of each peripheral area Ai for the target area Ai as the number of captured users Bi of the peripheral area Ai in the predetermined period T. User number calculation method and
For the target area Ai, it functions as a correlation learning engine in which a correlation model is pre-constructed with teacher data using the real estate utilization rate Ci as the objective variable and the number of users captured in the surrounding area Bi as the explanatory variable.
The correlation learning engine is characterized in that, for the target area Ai, the number of users captured in the surrounding area Bi is input, and the computer functions to estimate the occupancy rate Ci of the real estate.

According to other embodiments in the program of the invention
In order to estimate the number of staying users Ci currently staying in the target area Ai instead of the real estate occupancy rate Ci in the target area Ai,
Correlation learning engine
As a learning stage, for the target area Ai, a correlation model was constructed in advance using teacher data with the number of staying users Ci as the objective variable and the number of users captured in the surrounding area Bi as the explanatory variable.
As an estimation step, it is also preferable to input the number of users Bi captured in the surrounding area for the target area Ai and to make the computer function to estimate the number of staying users Ci currently staying.

According to other embodiments in the program of the invention
For the predetermined period T, the peripheral area allowable number of users storage means for storing the peripheral area allowable number of users in the entire peripheral area Aj (j = 1 to N) of the target area Ai, and the peripheral area allowable number of users storage means.
To further function as a means for calculating the peripheral area capture user rate Bi, which calculates the peripheral area capture user rate Bi of the number of peripheral area capture users Bi with respect to the allowable number of peripheral area users.
Correlation learning engine
As a learning stage, for the target area Ai, a correlation model is constructed in advance with teacher data using the real estate occupancy rate Ci as the objective variable and the surrounding area capture user rate Bi as the explanatory variable.
As an estimation stage, it is also preferable to input the surrounding area capture user rate Bi for the target area Ai and make the computer function to estimate the real estate occupancy rate Ci.

According to other embodiments in the program of the invention
It is also preferable to make the computer function so that the number of permissible users in the surrounding area is the permissible number of users or the number of rooms for real estate similar to the real estate in the target area Ai under predetermined conditions.

According to other embodiments in the program of the invention
A regression function derivation means for deriving a regression function f representing a transition tendency of the number of users Bi captured in the surrounding area in a plurality of predetermined periods T × m.
Further functions as a tendency difference calculation means for calculating the tendency difference DIFFi (= Bi－f (Bi)) between the current number of peripheral area capture users Bi and the predicted number of peripheral area capture users f (Bi) based on the regression function f. Let me
Correlation learning engine
As a learning stage, for the target area Ai, a correlation model is constructed in advance with teacher data using the real estate utilization rate Ci as the objective variable and the tendency difference DIFFi as the explanatory variable.
As an estimation stage, it is also preferable to input the tendency difference DIFFi for the target area Ai and make the computer function to estimate the occupancy rate Ci of the real estate.

According to other embodiments in the program of the invention
The weighting coefficient determining means uses the square root of the sum of squares of the differences between the plurality of characteristic quantities k as the degree of similarity d d (i, j) = √ (Σ _{k = 1} ^K (CHARi, k－CHARj, k) ² )
CHARi, k: Characteristic quantity k of region i
It is also preferable to make the computer function as such.

According to other embodiments in the program of the invention
The weighting coefficient determining means multiplies the degree of influence αk of each of the plurality of characteristic quantities k by the square of the difference of the characteristic quantities k, and sets the square root of the sum of the multiplied values as the degree of similarity d.
Similarity d (i, j) = √ (Σ _{k = 1} ^K ((CHARi, k－CHARj, k) ² × α k))
CHARi, k: Characteristic quantity k of region i
αk: It is also preferable to make the computer function according to the degree of influence of the characteristic quantity k.

According to other embodiments in the program of the invention
It is also preferable that the weighting coefficient determining means causes the computer to function so as to use the distance from the target area Ai to the surrounding area Aj as the similarity d between the characteristic amount of the target area Ai and the characteristic amount of the surrounding area Aj.

According to other embodiments in the program of the invention
The weighting coefficient determining means functions the computer to use the type, rank and / or evaluation of the real estate of the target area Ai and the surrounding area Aj as the similarity d between the characteristic amount of the target area Ai and the characteristic amount of the surrounding area Aj. It is also preferable to let them.

According to other embodiments in the program of the invention
The weighting coefficient determining means is a computer so as to use the number of floors of the building, the number of rooms, the room area, the number of employees and / or the usage fee as the similarity d between the characteristic amount of the target area Ai and the characteristic amount of the surrounding area Aj. It is also preferable to make it work.

According to other embodiments in the program of the invention
It is also preferable to make the computer function so that the occupancy rate Ci of the real estate as the correct answer value is the occupancy rate of the occupancy number with respect to the maximum capacity or the occupancy rate disclosed by the financial statement based on the real estate.

According to the present invention, it is a device for estimating the occupancy rate Ci of real estate in the target area Ai.
A user position database that accumulates time-series positions for each user who has a mobile terminal that can acquire the position,
A regional characteristic quantity storage means for storing one or more characteristic quantities of the region A for each region A,
Using the user location database, the number of captured users #A that stays in each region A for a predetermined period T is extracted, and the number of captured users is extracted.
Using the regional characteristic quantity storage means, each weighting coefficient ω according to the degree of similarity d between the characteristic quantity of the target area Ai and the characteristic quantity of a plurality of N surrounding areas Aj (j = 1 to N) with respect to the target area Ai. A weighting coefficient determining means for determining (Ai, Aj) and
Peripheral area capture calculated by weighting the weighting coefficient ω (Ai, Aj) to the number of captured users #Aj of each peripheral area Ai for the target area Ai as the number of captured users Bi of the peripheral area Ai in the predetermined period T. For the method of calculating the number of users and the target area Ai, it is made to function as a correlation learning engine in which a correlation model is constructed in advance with teacher data using the real estate utilization rate Ci as the objective variable and the number of users Bi captured in the surrounding area as the explanatory variable.
For the target area Ai, the number of users captured in the surrounding area Bi is input to the correlation learning engine, and the occupancy rate Ci of the real estate is estimated.

According to the present invention, it is an estimation method of an apparatus for estimating the occupancy rate Ci of real estate in the target area Ai.
The device is
A user position database that accumulates time-series positions for each user who has a mobile terminal that can acquire the position,
Each region A has a regional characteristic quantity storage unit that stores one or more characteristic quantities of the region A.
The device is
The first step of extracting the number of captured users #A staying in each region A for a predetermined period T using the user location database, and
Using the regional characteristic quantity storage unit, each weighting coefficient ω according to the degree of similarity d between the characteristic quantity of the target area Ai and the characteristic quantity of multiple N surrounding areas Aj (j = 1 to N) with respect to the target area Ai. The second step to determine (Ai, Aj) and
The sum of the number of captured users #Aj of each peripheral area Aj for the target area Ai weighted by the weighting coefficient ω (Ai, Aj) is calculated as the number of captured users Bi of the peripheral area Ai in the predetermined period T. As well as performing steps and
As a learning stage, a correlation model is constructed in advance using the correlation learning engine with teacher data for the target area Ai, with the real estate occupancy rate Ci as the objective variable and the number of users captured in the surrounding area Bi as the explanatory variable.
As an estimation stage, for the target area Ai, the number of users captured in the surrounding area Bi is input to the correlation learning engine, and the occupancy rate Ci of the real estate is estimated.

According to the program, device and method of the present invention, the occupancy rate of real estate in a relatively narrow target area can be estimated as accurately as possible even if the number of captured users is connected to the communication equipment of a specific telecommunications carrier. Can be done.

It is explanatory drawing which shows the number of acquired users in a specific telecommunications carrier for each region. It is a 1st functional block diagram of the estimation apparatus in this invention. It is explanatory drawing which shows the number of staying users of the surrounding area with respect to the target area. It is explanatory drawing which shows the number of captured users of the surrounding area with respect to the target area. It is explanatory drawing which calculates the number of peripheral area capture users based on the weighting coefficient ω. It is a 2nd functional block diagram of the estimation apparatus in this invention. It is a 3rd functional block diagram of the estimation apparatus in this invention. It is a 4th functional block diagram of the estimation apparatus in this invention. It is explanatory drawing which determines the influence degree α based on a weighting coefficient ω.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 2 is a first functional configuration diagram of the estimation device in the present invention.

The estimation device 1 estimates the occupancy rate Ci of the real estate in the target area Ai.
According to FIG. 2, the estimation device 1 is configured based on the correlation between the “occupancy rate Ci of the real estate in the target area Ai” and the “number of users captured in the surrounding area Ai in the target area Bi”. be.
The occupancy rate Ci of the real estate as the correct answer value is assumed to be the occupancy rate of the occupancy number with respect to the maximum capacity or the occupancy rate disclosed by the financial statement based on the real estate. This is based on historical data.

The estimation device 1 includes a user position database 11, a regional characteristic quantity storage unit 12, a capture user number extraction unit 13, a weight coefficient determination unit 14, a peripheral area capture user number calculation unit 15, and a correlation learning engine 10. Have. These functional components are realized as a program for operating the computer mounted on the device. In addition, the processing flow of these functional components can be understood as a method for estimating the real estate occupancy rate.

Here, the target area Ai for which the occupancy rate Ci is estimated and the surrounding area Aj are defined as follows.
Target area Ai: Set by the operator of the estimation work.
(It may be a specific real estate to be estimated)
Surrounding area Aj: Set in the specified area range including the target area Ai.
The surrounding area Aj may be within a predetermined radius centered on the target area Ai, or may be in units of prefectures, cities, towns and villages of the address.

FIG. 3 is an explanatory diagram showing the number of users staying in the surrounding area with respect to the target area.

According to FIG. 3, the surrounding areas A2 to A8 exist within the predetermined area with respect to the target area A1. Specifically, other hotels A2 to A8 exist within a predetermined radius centering on the position of the real estate in the target area A1 in the hotel. It is considered that the tendency of the number of staying users in the real estate of the target area A1 changes in the same way as the number of staying users in the real estate of the neighboring surrounding area Aj of the same type.

[User location database 11]
The user position database 11 accumulates time-series positions for each user who has a mobile terminal capable of acquiring a position.
According to FIG. 2, the user location database 11 is operated and managed by a specific telecommunications carrier, and the time and location information are stored in association with each subscriber ID (user ID).
Here, as an important point, the user position database 11 is a captured user by the telecommunications business equipment of a specific telecommunications carrier, and is not all actual users. That is, the location information is collected only for the user who holds the mobile terminal 2 contracted with a specific telecommunications carrier.

The positions accumulated in the user position database 11 are, for example, as follows.
(1) Terminal positioning position positioned by the mobile terminal 2 possessed by the user This is latitude / longitude information positioned by the mobile terminal 2 itself by GPS (Global Positioning System).
(2) Base station positioning position of a mobile terminal connected to a communication carrier's base station or access point The position of the mobile terminal 2 is estimated from the position information of the base station or access point under the mobile terminal 2. You may. However, this position information has a coarse spatial particle size.
These position information may be represented by latitude / longitude or map coordinates, or may be converted into an address name or a map mesh number.

[Regional characteristic quantity storage unit 12]
The regional characteristic quantity storage unit 12 stores one or more characteristic quantities of the region A together with the region range for each region A.
Examples of the "characteristic quantity" include the following.
Location of real estate in target area A Type, rank and / or evaluation of real estate in target area A

The location of the real estate is used, for example, to calculate the distance between the area Ai and the area Aj.
The type of real estate may be the number of floors, the number of rooms, the room area, the number of employees, and the usage fee in the real estate of the target area A. For example, in the case of a hotel, it may be a hotel mode (luxury, business, guest house, capsule, etc.) or accommodation fee.
The rank of the real estate may be, for example, in the case of a hotel, a class (such as 4 stars or 5 stars).
The evaluation of the real estate may be, for example, the evaluation of the user by the SNS (Social Networking Service).

[Captured user number extraction unit 13]
The captured user number extraction unit 13 extracts the captured user number # A (T) staying in each region A for a predetermined period T by using the user position database 11. As shown in FIG. 1 described above, the number of captured users #A (T) connected to the telecommunications business equipment of a specific telecommunications carrier is extracted for each region A.

Here, the term "region" means that the more the flat mesh on the map is subdivided, the more real estate such as buildings and facilities can be specified. Further, the "region" may be divided into multidimensional spaces such as the number of floors of the building or facility. For example, when dividing in three dimensions, the space can also be called a "cube".

FIG. 4 is an explanatory diagram showing the number of captured users in the surrounding area with respect to the target area.

According to FIG. 4, as compared with FIG. 3, only the number of captured users connected to the communication business equipment of a specific communication carrier is shown. The number of captured users in the real estate hotel in the target area A1 and the number of captured users in other hotels A2 to A8 are shown.

[Weighting factor determination unit 14]
The weighting coefficient determination unit 14 uses the regional characteristic quantity storage unit 12 to determine the degree of similarity between the characteristic quantity of the target area Ai and the characteristic quantity of a plurality of N surrounding areas Aj (j = 1 to N) with respect to the target area Ai. Each weighting coefficient ω (Ai, Aj) corresponding to d is determined.
The weighting coefficient ω may be variable depending on the period T, or may be ω (Ai, Aj, T).

The weighting coefficient ω is determined in the following embodiment, for example, according to the characteristic quantity.
(When the characteristic quantity is the position of the real estate)
The weighting coefficient determining unit 14 uses the "distance" from the target area Ai to the surrounding area Aj as the degree of similarity d between the characteristic amount of the target area Ai and the characteristic amount of the surrounding area Aj. The distance here is a straight line distance for expressing physical closeness and distance, which can be calculated from the difference between longitude and latitude.
Distance: d (Ai, Aj)
Weight coefficient: ω (Ai, Aj) = 1 / (1 + d (Ai, Aj))
The surrounding area Aj also includes the same area (Ai = Aj) as the target area Ai. When i = j, d (Ai, Aj) = 0.

(When the characteristic quantity is the type, rank and / or evaluation of real estate)
The weighting coefficient determining unit 14 uses the “type, rank and / or evaluation” of the real estate in the target area Ai as the similarity d between the characteristic amount of the target area Ai and the characteristic amount of the surrounding area Aj.
For example, when the real estate in the target area Ai is ranked “☆☆”, the weighting coefficient of the real estate in the surrounding area Aj is as follows.
Target area Ai real estate: Rank "☆☆"
Surrounding area Aj real estate: In the case of rank "☆☆☆"
d (Ai, Aj) = 1 (difference between ☆☆ and ☆☆☆ = 1)
ω (Ai, Aj) = 1 / (1 + 1) = 0.5
Surrounding area Aj real estate: In the case of rank "☆☆"
d (Ai, Aj) = 1 (difference between ☆☆ and ☆☆ = 0)
ω (Ai, Aj) = 1 / (1 + 0) = 1
Real estate in the surrounding area Aj: In the case of rank "☆"
d (Ai, Aj) = 1 (difference between ☆☆ and ☆☆☆ 1)
ω (Ai, Aj) = 1 / (1 + 1) = 0.5

The weighting coefficient determining unit 14 may calculate the similarity d based on a plurality of characteristic quantities k (for example, mode, rank, evaluation, etc.). In the simplest case, the square root of the sum of squares of the differences between the characteristic quantities k may be set as the similarity d as follows.
d (i, j) = √ (Σ _{k = 1} ^K (CHARi, k－CHARj, k) ² )
CHARi, k: Characteristic quantity k of region i

[Peripheral area capture user number calculation unit 15]
The peripheral area capture user number calculation unit 15 calculates the sum of the number of captured users #Aj of each peripheral area Aj with respect to the target area Ai weighted by the weighting coefficient ω (Ai, Aj), and the peripheral area of the target area Ai in the predetermined period T. Calculated as the number of captured users Bi.
Bi (T) ＝ Σ _{j = 1} ^N (#Aj (T) × ω (Ai, Aj))
= Σ _{j = 1} ^N (#Aj (T) × (1 / (1 + d (Ai, Aj))))

According to the present invention, the occupancy rate C1 of the real estate in the target area A1 is estimated based on the number of captured users of the real estate in the neighboring surrounding area Aj. That is, even if the estimated number of captured users for one real estate varies, the population is expanded by estimating the number of captured users that has expanded to the surrounding real estate of the same type.

FIG. 5 is an explanatory diagram for calculating the number of users who capture the surrounding area based on the weighting coefficient ω.

According to FIGS. 4 and 5, real estate in the surrounding area Aj, such as a hotel, is ranked as follows.
Real estate in target area A1: Rank "☆☆"
Real estate in surrounding area A2: Rank "☆"
Surrounding area A3 real estate: Rank "☆☆"
Surrounding area A4 real estate: Rank "☆☆☆"
Surrounding area A5 real estate: Rank "☆☆"
Surrounding area A6 real estate: Rank "☆☆☆"
Surrounding area A7 real estate: Rank "☆"
Surrounding area A8 real estate: Rank "☆☆☆"
In this case, the number of users B1 captured in the surrounding area of the target area A1 is calculated as follows.
B1 ＝ # A3 × ω (A1, A3) ＋ # A5 × ω (A1, A5)
＋ # A4 × ω (A1, A4) ＋ # A6 × ω (A1, A6) ＋ # A8 × ω (A1, A8)
＋ # A2 × ω (A1, A2) ＋ # A7 × ω (A1, A7)
The number of captured users B1 in the surrounding area of the target area A1 calculated here is strongly influenced by the number of captured users in the peripheral area having a high degree of similarity d between the target area A1 and the characteristic quantity, and conversely, the characteristic quantity with the target area A1. It is less likely to be affected by the number of captured users in the surrounding area where the similarity d of is low.

[First correlation learning engine 10]
The first correlation learning engine 10 estimates the occupancy rate Ci of the real estate as an estimation step.
The first correlation learning engine 10 constructs a correlation model in advance for the target area Ai based on the following teacher data as a learning stage.
Objective variable: Real estate occupancy rate Ci
Explanatory variable: Number of users captured in the surrounding area Bi
The number of peripheral area capture users Bi input as teacher data in the learning stage of the first correlation learning engine 10 is calculated by the peripheral area capture user number calculation unit 15 in the past.

The correlation model is constructed so that the correlation is maximized between the real estate occupancy rate Ci and the number of users captured in the surrounding area Bi.
Here, the real estate occupancy rate Ci and the number of users captured in the surrounding area Bi as teacher data are counted in the past. For example, when the occupancy rate C1 of the real estate 1 is based on the information of the financial statements of the real estate 1, it is associated with the number of users Bi for capturing the surrounding area calculated by the estimation device 1 at the same time.
If the real estate is a hotel for accommodation, the occupancy rate Ci of the real estate may be the ratio of the number of rooms actually stayed to the total number of accommodation rooms.

Then, the first correlation learning engine 10 inputs the peripheral area capture user number Bi calculated by the peripheral area capture user number calculation unit 15 for the target area Ai as the estimation step. On the other hand, the occupancy rate Ci of the real estate is output.

FIG. 6 is a second functional configuration diagram of the estimation device in the present invention.

According to FIG. 6, the estimation device 1 estimates the number of staying users Ci of the real estate in the target area Ai.
The estimation device 1 is configured based on the correlation between "the number of users staying in the real estate of the target area Ai Ci" and "the number of users capturing the surrounding area of the target area Ai Bi".
The number of staying users of the real estate as the correct answer Ci is, for example, the number of users counted in the real estate or the number of staying users disclosed in the financial statements based on the real estate. This is based on historical data.

According to FIG. 6, only the correlation learning engine 10 is different from that of FIG.

[Second correlation learning engine 10]
The second correlation learning engine 10 estimates the number of staying users Ci currently staying in the target area Ai as an estimation step.
The second correlation learning engine 10 constructs a correlation model in advance with the following teacher data for the target area Ai as a learning stage.
Objective variable: Number of staying users Ci
Explanatory variable: Number of users captured in the surrounding area Bi
The number of peripheral area capture users Bi input as teacher data in the learning stage of the second correlation learning engine 10 is calculated by the peripheral area capture user number calculation unit 15 in the past.

The correlation model is constructed so that the correlation is maximized between the number of staying users Ci of the real estate and the number of surrounding area capture users Bi.
Here, the number of staying users Ci and the number of surrounding area capture users Bi staying in the real estate as teacher data are counted in the past.

Then, the second correlation learning engine 10 inputs the peripheral area capture user number Bi calculated by the peripheral area capture user number calculation unit 15 for the target area Ai as the estimation step. On the other hand, the number of staying users Ci who stayed in the real estate is output.

FIG. 7 is a third functional configuration diagram of the estimation device in the present invention.

According to FIG. 7, the estimation device 1 estimates the occupancy rate Ci of the real estate in the target area Ai.
The estimation device 1 is configured based on the correlation between "the occupancy rate Ci of the real estate in the target area Ai" and "the number of users captured in the surrounding area Ai in the target area Bi".

According to FIG. 7, as compared with FIG. 2, the peripheral area allowable number of users storage unit 16 and the peripheral area capture user rate calculation unit 17 are further provided, and the correlation learning engine 10 is different.

[Area area allowable number of users storage unit 16]
The peripheral area allowable number storage unit 16 stores the number of peripheral area allowable users in the entire peripheral area Aj (j = 1 to N) of the target area Ai for the predetermined period T.
The permissible number of users in the surrounding area is the number of users based on the permissible number of rooms for real estate similar to the real estate in the target area Ai under predetermined conditions. For example, when the real estate in the surrounding area Aj is a hotel, the allowable number of users in the surrounding area may be the number of users based on the number of accommodation rooms in the hotel in the surrounding area Aj.

[Peripheral area capture user rate calculation unit 17]
The peripheral area capture user rate calculation unit 17 calculates the peripheral area capture user rate Bi of the peripheral area capture user number Bi with respect to the peripheral area allowable number of users.
Peripheral area capture user rate Bi = Number of peripheral area capture users Bi
/ Permissible number of users in the surrounding area

[Third correlation learning engine 10]
The third correlation learning engine estimates the occupancy rate Ci of the real estate as an estimation step.
The third correlation learning engine 10 constructs a correlation model in advance with the following teacher data for the target area Ai as a learning stage.
Objective variable: Real estate occupancy rate Ci
Explanatory variable: Surrounding area capture user rate Bi
The peripheral area capture user rate Bi input as teacher data in the learning stage of the third correlation learning engine 10 is calculated by the peripheral area capture user rate calculation unit 17 in the past.
The correlation model is constructed so that the correlation is maximized between the real estate occupancy rate Ci and the surrounding area capture user rate Bi.

As the estimation step, the third correlation learning engine inputs the peripheral area capture user rate Bi calculated by the peripheral area capture user number calculation rate 17 for the target area Ai. Then, the occupancy rate Ci of the real estate is estimated.

FIG. 8 is a fourth functional configuration diagram of the estimation device in the present invention.

According to FIG. 8, the estimation device 1 estimates the occupancy rate Ci of the real estate in the target area Ai.
The estimation device 1 is configured based on the correlation between the “occupancy rate Ci of the real estate in the target area Ai” and the “trend difference DIFFi in the target area Ai”.

According to FIG. 8, as compared with FIG. 2, the regression function derivation unit 18, the tendency difference calculation unit 19, and the tendency difference are further provided, and the correlation learning engine 10 is different.

[Regression function derivation unit 18]
The regression function derivation unit 18 derives a regression function f that represents a transition tendency of the number of users Bi for capturing the surrounding area in a plurality of predetermined periods T × m.
The regression function f represents the difference in the change (trend) in the number of users captured in the surrounding area Bi, that is, the change in the entire population over a long period of time. As the regression model, for example, it can be expressed by the following linear equation, but it is not limited to it.
f (Bi (T)) = a × Bi (T) + ba: tilt, b: intercept

[Trend difference calculation unit 19]
The tendency difference calculation unit 19 calculates the tendency difference DIFFi between the current number of peripheral area capture users Bi and the predicted number of peripheral area capture users f (Bi) based on the regression function f.
DIFFi ＝ Bi－f (Bi)

FIG. 9 is an explanatory diagram for determining the degree of influence α based on the weighting coefficient ω.

The weighting coefficient determining unit 14 may be obtained by adding the squares of the differences of the characteristic quantities k according to the influence degree αk of each of the plurality of characteristic quantities k. That is, the characteristic amount k showing a strong influence degree α is weighted, and the characteristic amount k having the lowest influence degree is adjusted to be “0”.
The weighting coefficient determination unit 14 multiplies the influence degree αk of each of the plurality of characteristic quantities k (for example, mode, rank, evaluation, etc.) by the square of the difference of the characteristic quantities k, and sets the square root of the sum of the multiplied values to the similarity degree d. And.
Similarity d (i, j) = √ (Σ _{k = 1} ^K ((CHARi, k－CHARj, k) ² × α k))
CHARi, k: Characteristic quantity k of region i
αk: Degree of influence of characteristic quantity k According to FIG. 9, the degree of similarity between region A1 and region A4 is calculated as follows, for example.
d (1,4) = √ ((Rank difference ² / Impact α ₁ ) + (Distance ² / Impact α ₂ )
＋ (Difference in the number of rooms ² / Impact α ₃ ) ＋ (Difference in the number of employees ² / Impact α ₄ )
= √ ((2-3) ² α ₁ ＋ (0-1) ² α ₂ ＋ (100-10) ² α ₃ ＋ (30-28) ² α ₄ )
Then, the weighting coefficient ω is calculated as follows from the similarity d in consideration of the influence degree α for each characteristic quantity k.
ω (i, j) = 1 / (1 + d (i, j))

Next, the peripheral area capture user number calculation unit 15 can calculate the peripheral area capture user number Bi as follows using the weighting coefficient ω.
Bi = Σ _{j = 1} ^N (#Aj × ω (i, j))

Next, the tendency difference calculation unit 19 calculates the tendency difference DIFFi of the target area i from the peripheral area capture user number Bi using the regression function derivation unit 18.

According to FIG. 9, it further has an influence degree determination unit 141. The influence degree determination unit 141 calculates the influence degree αk that maximizes the correlation between the operating rate Ci and the tendency difference DIFFi.
argmax α ρ (Ci, DIFFi)
Function ρ: A function that calculates the correlation coefficient between the operating rate Ci and the tendency difference DIFFi. d can be calculated, and the weighting coefficient ω can be determined by the degree of similarity d.
The degree of influence α has the effect of standardizing the characteristic quantity and the effect of differentiating it from other characteristic quantities by adding the strength and weakness of the substantial characteristic quantity. As a result, the estimation accuracy of the correlation coefficient with the operating rate Ci can be improved.

[Fourth Correlation Learning Engine 10]
The fourth correlation learning engine 10 estimates the occupancy rate Ci of the real estate as an estimation step.
The fourth correlation learning engine 10 constructs a correlation model in advance with the following teacher data for the target area Ai as a learning stage.
Objective variable: Real estate occupancy rate Ci
Explanatory variable: Trend difference DIFFi
The tendency difference DIFFi input as teacher data in the learning stage of the fourth correlation learning engine 10 has been calculated by the tendency difference calculation unit 19 in the past.
That is, the correlation model is constructed so that the correlation is maximized between the real estate occupancy rate Ci and the trend difference DIFFi.

The fourth correlation learning engine inputs the tendency difference DIFFi calculated by the tendency difference calculation unit 19 for the target area Ai as the estimation step, and estimates the occupancy rate Ci of the real estate.

The first to fourth correlation learning engines 10 described above may be any learning engine for linear regression. In order to avoid overfitting, it is also necessary to optimize the teacher data and verify the model accuracy during operation, using the expected log-likelihood as a measure for the number of parameters.
For example, the parameter that maximizes the correlation may be set to zero by partially differentiating the objective variable with each parameter, and the simultaneous equations may be solved or the parameter that maximizes the correlation coefficient may be approximated.

As another embodiment, the gradation of the hue is darkened (or lightened) for each region according to the occupancy rate Ci (or the number of staying users Ci) of the real estate estimated by the first to fourth correlation learning engines 10. You can also create a displayed real estate occupancy rate map. You can recognize the real estate occupancy rate Ci etc. just by looking at the color of the real estate occupancy rate map, and it will be easier to judge the property value of the real estate in that area.

According to the application of the present invention, for example, the following real estate can be assumed by dividing the area.
(1) Business establishment type: For example, company building or factory (2) Residence type: For example, apartment or residential area (3) Commercial facility type: For example, shopping mall or pachinko store (4) Hotel type: For example, accommodation facility Conference hall (5) Public facility-based type: For example, library (6) Healthcare facility-based type: For example, sports gym (7) (1)-(6) integrated / complex type (8) "Company" that uses the real estate Value Judgment: Can be used, for example, when acquiring a company

As described in detail above, according to the program, apparatus and method of the present invention, the occupancy rate of real estate in a relatively narrow target area is achieved even if the number of captured users is connected to the communication equipment of a specific telecommunications carrier. Can be estimated as accurately as possible.

Further, according to the present invention, real estate investors and related parties can know the occupancy rate of real estate without waiting for the announcement of the number of users for each accounting period. It is possible to manage the risk of investment based on the property value without having to carry the risk of fluctuations in the occupancy rate of real estate.

Various changes, modifications and omissions of the technical idea and the scope of view of the present invention can be easily carried out by those skilled in the art with respect to the various embodiments of the present invention described above. The above explanation is just an example and does not attempt to limit anything. The present invention is limited only to the scope of claims and their equivalents.

1 Estimator 10 Correlation learning engine 11 User position database 12 Regional characteristic quantity storage unit 13 Captured user number extraction unit 14 Weight coefficient determination unit 141 Impact degree determination unit 15 Peripheral area capture user number calculation unit 16 Peripheral area allowable number of users storage unit 17 Peripheral area capture user rate calculation unit 18 Regression function derivation unit 19 Trend difference calculation unit 2 Mobile terminal

Claims (13)

It is a program that makes the computer installed in the device that estimates the occupancy rate Ci of the real estate in the target area Ai function.
A user position database that accumulates time-series positions for each user who has a mobile terminal that can acquire the position,
A regional characteristic quantity storage means for storing one or more characteristic quantities of the region A for each region A,
Using the user location database, the number of captured users #A that stays in each region A for a predetermined period T is extracted, and the number of captured users is extracted.
Using the regional characteristic quantity storage means, each weighting coefficient ω according to the degree of similarity d between the characteristic quantity of the target area Ai and the characteristic quantity of a plurality of N surrounding areas Aj (j = 1 to N) with respect to the target area Ai. A weighting coefficient determining means for determining (Ai, Aj) and
Peripheral area capture calculated by weighting the weighting coefficient ω (Ai, Aj) to the number of captured users #Aj of each peripheral area Ai for the target area Ai as the number of captured users Bi of the peripheral area Ai in the predetermined period T. User number calculation method and
For the target area Ai, it functions as a correlation learning engine in which a correlation model is pre-constructed with teacher data using the real estate utilization rate Ci as the objective variable and the number of users captured in the surrounding area Bi as the explanatory variable.
The correlation learning engine is a program characterized by inputting the number of users captured in the surrounding area Bi for the target area Ai and making the computer function to estimate the occupancy rate Ci of the real estate. In order to estimate the number of staying users Ci currently staying in the target area Ai instead of the real estate occupancy rate Ci in the target area Ai,
Correlation learning engine
As a learning stage, for the target area Ai, a correlation model was constructed in advance using teacher data with the number of staying users Ci as the objective variable and the number of users captured in the surrounding area Bi as the explanatory variable.
The program according to claim 1, wherein, as an estimation step, the computer functions to input the number of users Bi captured in the surrounding area and estimate the number of staying users Ci currently staying in the target area Ai. For the predetermined period T, the peripheral area allowable number of users storage means for storing the peripheral area allowable number of users in the entire peripheral area Aj (j = 1 to N) of the target area Ai, and the peripheral area allowable number of users storage means.
To further function as a means for calculating the peripheral area capture user rate Bi, which calculates the peripheral area capture user rate Bi of the number of peripheral area capture users Bi with respect to the allowable number of peripheral area users.
Correlation learning engine
As a learning stage, for the target area Ai, a correlation model is constructed in advance with teacher data using the real estate occupancy rate Ci as the objective variable and the surrounding area capture user rate Bi as the explanatory variable.
The program according to claim 1, wherein, as an estimation stage, the computer is made to function so as to input the surrounding area capture user rate Bi for the target area Ai and estimate the real estate occupancy rate Ci. The program according to claim 3, wherein the permissible number of users in the surrounding area is characterized in that the computer functions so that the number of users or the number of rooms is permissible for real estate similar to the real estate in the target area Ai under predetermined conditions. A regression function derivation means for deriving a regression function f representing a transition tendency of the number of users Bi captured in the surrounding area in a plurality of predetermined periods T × m.
Further functions as a tendency difference calculation means for calculating the tendency difference DIFFi (= Bi－f (Bi)) between the current number of peripheral area capture users Bi and the predicted number of peripheral area capture users f (Bi) based on the regression function f. Let me
Correlation learning engine
As a learning stage, for the target area Ai, a correlation model is constructed in advance with teacher data using the real estate utilization rate Ci as the objective variable and the tendency difference DIFFi as the explanatory variable.
The program according to claim 1, wherein, as an estimation stage, the tendency difference DIFFi is input for the target area Ai, and the computer functions to estimate the occupancy rate Ci of the real estate. The weighting coefficient determining means uses the square root of the sum of squares of the differences between the plurality of characteristic quantities k as the degree of similarity d d (i, j) = √ (Σ _{k = 1} ^K (CHARi, k－CHARj, k) ² )
CHARi, k: Characteristic quantity k of region i
The program according to any one of claims 1 to 5, wherein the computer functions as described above. The weighting coefficient determining means multiplies the degree of influence αk of each of the plurality of characteristic quantities k by the square of the difference of the characteristic quantities k, and sets the square root of the sum of the multiplied values as the degree of similarity d.
Similarity d (i, j) = √ (Σ _{k = 1} ^K ((CHARi, k－CHARj, k) ² × α k))
CHARi, k: Characteristic quantity k of region i
αk: The program according to claim 5, wherein the computer functions as the degree of influence of the characteristic quantity k. The weighting coefficient determining means is characterized in that the computer functions to use the distance from the target area Ai to the surrounding area Aj as the similarity d between the characteristic amount of the target area Ai and the characteristic amount of the surrounding area Aj. The program according to any one of Items 1 to 7. The weighting coefficient determining means functions the computer to use the type, rank and / or evaluation of the real estate of the target area Ai and the surrounding area Aj as the similarity d between the characteristic amount of the target area Ai and the characteristic amount of the surrounding area Aj. The program according to any one of claims 1 to 7, wherein the program is to be used. The weighting coefficient determining means is a computer so as to use the number of floors of the building, the number of rooms, the room area, the number of employees and / or the usage fee as the similarity d between the characteristic amount of the target area Ai and the characteristic amount of the surrounding area Aj. The program according to claim 9, wherein the program is made to function. The occupancy rate Ci of the real estate as the correct answer value is a claim characterized by making the computer function so as to be the occupancy rate of the occupancy number with respect to the maximum capacity or the occupancy rate disclosed by the financial statement based on the real estate. Item 1. The program according to any one of items 1 to 10. A device that estimates the occupancy rate Ci of real estate in the target area Ai.
A user position database that accumulates time-series positions for each user who has a mobile terminal that can acquire the position,
A regional characteristic quantity storage means for storing one or more characteristic quantities of the region A for each region A,
Using the user location database, the number of captured users #A that stays in each region A for a predetermined period T is extracted, and the number of captured users is extracted.
Using the regional characteristic quantity storage means, each weighting coefficient ω according to the degree of similarity d between the characteristic quantity of the target area Ai and the characteristic quantity of a plurality of N surrounding areas Aj (j = 1 to N) with respect to the target area Ai. A weighting coefficient determining means for determining (Ai, Aj) and
Peripheral area capture calculated by weighting the weighting coefficient ω (Ai, Aj) to the number of captured users #Aj of each peripheral area Ai for the target area Ai as the number of captured users Bi of the peripheral area Ai in the predetermined period T. For the method of calculating the number of users and the target area Ai, it is made to function as a correlation learning engine in which a correlation model is constructed in advance with teacher data using the real estate utilization rate Ci as the objective variable and the number of users Bi captured in the surrounding area as the explanatory variable.
A device characterized by inputting the number of users Bi captured in the surrounding area into the correlation learning engine for the target area Ai and estimating the occupancy rate Ci of the real estate. It is an estimation method of a device that estimates the occupancy rate Ci of real estate in the target area Ai.
The device is
A user position database that accumulates time-series positions for each user who has a mobile terminal that can acquire the position,
Each region A has a regional characteristic quantity storage unit that stores one or more characteristic quantities of the region A.
The device is
The first step of extracting the number of captured users #A staying in each region A for a predetermined period T using the user location database, and
Using the regional characteristic quantity storage unit, each weighting coefficient ω according to the degree of similarity d between the characteristic quantity of the target area Ai and the characteristic quantity of multiple N surrounding areas Aj (j = 1 to N) with respect to the target area Ai. The second step to determine (Ai, Aj) and
The sum of the number of captured users #Aj of each peripheral area Aj for the target area Ai weighted by the weighting coefficient ω (Ai, Aj) is calculated as the number of captured users Bi of the peripheral area Ai in the predetermined period T. As well as performing steps and
As a learning stage, a correlation model is constructed in advance using the correlation learning engine with teacher data for the target area Ai, with the real estate occupancy rate Ci as the objective variable and the number of users captured in the surrounding area Bi as the explanatory variable.
As an estimation stage, an estimation method characterized by inputting the number of users captured in the surrounding area Bi into the correlation learning engine for the target area Ai and estimating the occupancy rate Ci of the real estate.

Images