JP7288558B1 - Prediction device and program - Google Patents

Abstract

Kind Code: A1 A real estate purchase or brokerage business can be conducted for an owner at a stage before a property to be sold becomes actualized. SOLUTION: An attribute information acquisition unit 12 for acquiring attribute information of a customer, a lifeline information acquisition unit 13 for acquiring information on the customer's lifeline, and suspension information for acquiring information on suspension of use of the customer's lifeline. The acquisition unit 14 performs machine learning on the attribute of the customer who acquired the information on the suspension of use and the information on the past use of the lifeline as training data, so that the acquired attribute of the other customer and the information on the lifeline can be used to identify the other customer. and a prediction unit 15 for predicting suspension of use of the lifeline. [Selection drawing] Fig. 3

Description

The present invention relates to a prediction device and program.

As awareness of environmental problems and the effective use of resources is increasing, the tendency of overemphasis on new construction of real estate is being reconsidered, and the second-hand housing market such as second-hand condominiums is being activated.
For example, Patent Literature 1 discloses a real estate purchase system connected to a network that supports the purchase of real estate. A property registration means for registering property information and a property provision means for preparing property provision information to be provided on the network from the property information registered by the property registration means, wherein the network communication means is prepared by the property provision means. A technique for disclosing property provision information obtained through a network is described.

JP-A-2002-024596

Here, in recent years, in addition to inter-individual transactions in which second-hand properties are bought and sold via brokers, second-hand properties are purchased by brokers, remodeled and renovated to increase asset value, and then resold at a lower price than new construction. You have a reseller business. In such a purchase and resale business, it is important from the standpoint of business continuity to purchase a large amount of second-hand properties, as in the conventional brokerage business.
For owners of second-hand properties who are considering selling, in general, real estate agents (brokers, purchase resellers, (regardless of whether)), the owner himself contacts and conveys his intention to sell himself, but the approach method from the real estate agent to the owner is limited as mentioned above, and they will wait passively. In other words, it is difficult for real estate agents to identify owners who intend to sell.
It is assumed that a person who intends to sell his or her real estate as a second-hand property has already applied for suspension of use of the lifeline at the time of notifying the real estate agent of the intention to sell. In such a case, the real estate agent, or if the lifeline operator whose use has been suspended is developing a business related to the used property, the said lifeline operator will be in an advantageous position to handle the used property compared to other operators. One thing makes it difficult for other dealers to continue to stock such second-hand properties. Business related to second-hand properties includes, for example, moving business, remodeling business, etc., in addition to buying and selling and intermediation.

The purpose of the present invention is to solve such problems. Specifically, it is possible to conduct real estate purchase or brokerage business for owners before the property for sale becomes apparent. It is to be.

Based on this object, the present invention provides an attribute information acquisition unit that acquires customer attribute information, a lifeline information acquisition unit that acquires information on the customer's lifeline, and information on the suspension of use of the customer's lifeline. By machine-learning the acquisition unit for suspension of use information to be acquired, the attributes of the customer who acquired the information on suspension of use, and the information on the past use of the lifeline as training data, the attributes of other customers and the information on the lifeline acquired and a prediction unit that predicts suspension of use of the other customer's lifeline from information.
Here, it is possible to machine-learn the information regarding the customer's phone call as the training data by the prediction unit when the information regarding the suspension of use of the customer's lifeline is acquired. Further, the information on the customer's lifeline can include information indicating the medium type of the contact history with the customer. Also, the information on the customer's lifeline can include information indicating whether the telephone number is a fixed telephone number.
Further, the information on the customer's lifeline can include information indicating whether or not the telephone number is registered as the customer's attribute information.
Further, the information on the customer's lifeline can include information indicating when, who, for what purpose, with what medium, and what was applied as a customer contact history.
Furthermore, the information on the lifeline can include information indicating an elapsed period from the start of use.
Furthermore, in the information on the lifeline, the ratio of the usage amount in a predetermined usage period to the usage amount in a predetermined time period in a day is calculated for each customer with respect to the average value of a plurality of customers. Information indicating the results of the comparison may be included.
Furthermore, the information about the lifeline can include information indicating that there is a change in the usage amount in a predetermined usage period and the latest usage amount trend.
According to another aspect of the present invention to achieve the above objects, an information processing apparatus includes an attribute information acquisition function for acquiring customer attribute information, a lifeline information acquisition function for acquiring information relating to the customer's lifeline, and a lifeline information acquisition function for acquiring the customer's lifeline information. Use suspension information acquisition function that acquires information on suspension of use of the line, attributes of the customer who acquired information on suspension of use, and information on past use of the lifeline as training data. and a prediction function for predicting suspension of use of the other customer's lifeline from the customer's attributes and information on the lifeline.

According to the present invention, it is possible to conduct real estate purchase or brokerage business to the owner before the property for sale becomes actualized.

BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows the whole structural example of the real estate transaction guidance system which concerns on this embodiment. It is a figure which shows the hardware configuration example of a prediction apparatus and a terminal. It is a figure which shows the functional structural example of a prediction apparatus. It is a figure explaining a machine-learning model. FIG. 10 is a diagram for explaining the extraction of customers whose gas consumption has changed, (a) and (c) are the monthly consumption until March, and (b) is the same customer's monthly consumption last year. indicates FIG. 10 is a diagram illustrating another example of training data; FIG. 10 is a diagram showing an example of a screen output and displayed on a terminal of a business operator in the case of telephone reception, (a) shows an example of a reception screen for a closing request for a telephone operator, and (b) is another reception screen. An example is partially shown, and (c) partially shows another example of the reception screen. FIG. 10 is a diagram showing an example of a screen output and displayed on a user's terminal in the case of Web reception, (a) shows a Web reception screen (1/n) for a closing application, and (b) shows a Web reception screen for a closing application. An example of a reception screen (2/n) is shown. FIG. 10 is a diagram showing an example of a screen output and displayed on a user's terminal in the case of Web reception, (a) shows another example of the reception screen (2/n), and (b) shows the reception screen (2 /n).

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
FIG. 1 is a diagram showing an example of the overall configuration of a real estate transaction guidance system 1 according to this embodiment.
A real estate transaction guidance system 1 according to the present embodiment estimates vacant houses that will no longer be lived in by using information on contracts or usage of infrastructure projects, and guides candidates for real estate transaction partners that improve the operating efficiency of real estate transactions. is.
Real estate presumed to be vacant includes cases where the current residents are planning to move, and cases where they have already moved.

The infrastructure business referred to here refers to business as a foundation that supports industry and life, and includes energy infrastructure business such as electricity and gas. Changes in contracts and usage of energy infrastructure projects are useful information for estimating vacant houses. Also, the infrastructure projects referred to herein may include other infrastructure projects that may be useful for vacant house estimation. Although there are various classifications of infrastructure business, for example, life infrastructure business such as communication infrastructure such as telephone network and communication network can be mentioned in addition to water supply and sewerage and broadcast reception.

The real estate transaction guidance system 1 includes a prediction device 10, a terminal 20, a terminal 30, and a terminal 40. Prediction device 10 and terminal 20 , prediction device 10 and terminal 30 , and prediction device 10 and terminal 40 are connected via network 90 .

The terminal 20 is a device of a consumer, user, or user (hereinafter referred to as a user) related to infrastructure business, and is located in a residential building together with the telephone T20.
The terminal 30 is a device of an infrastructure business supplier, provider, or operator (hereinafter referred to as an operator). The operator has a telephone T30 for communicating with the user's telephone T20.
The terminal 40 is a device of a trader who conducts real estate transactions, such as real estate purchase transactions and real estate brokerage transactions. The trader has a printer P40 for performing various printing from the terminal 40. FIG.

The prediction device 10 of the real estate transaction guidance system 1 acquires information on second-hand houses such as second-hand condominiums, and also acquires information on contracts and usage conditions for gas, electric power, water supply, telephone, communication, etc. in the second-hand houses. More specifically, the information on second-hand houses described above may be obtained from a database of housing information or the like, or may be obtained automatically or manually from information published on the Internet or the like.
Pre-owned homes here are residential buildings that are currently occupied by residents, and do not include residential buildings that have not been occupied since the time of construction. To explain further, pre-owned houses include cases where the current resident has lived in the house since the time of construction, and cases where there was a previous resident before the current resident. Also, a newly built condominium for which a sales contract has not been concluded even after a predetermined number of years have passed since construction is called a new and old condominium, but such a new and old condominium is not included in second-hand housing.
In addition, second-hand properties include housing complexes such as condominiums as residential buildings, as well as detached houses for sale. In the case of a condominium, it is a room, which is a unit for sale.

The prediction device 10 is implemented by, for example, an information processing device such as a computer. The prediction device 10 may be configured by a single computer, or may be implemented by distributed processing by a plurality of computers.

As described above, the terminal 20 is a user's device for the infrastructure business, and is implemented by, for example, a PC (Personal Computer).
As described above, the terminal 30 is a device of an infrastructure business operator, and is realized by, for example, a PC. Businesses include, for example, a gas business that supplies gas and an electric power business that supplies electric power. Such gas companies and power companies can be called service providers.
The terminal 40 is, as described above, a device of a real estate transaction dealer, and is realized by, for example, a PC. Traders include real estate purchasers and real estate brokers.

In the present embodiment, a configuration example in which the terminal 30 and the terminal 40 are devices different from each other is adopted. . Further, in the present embodiment, a configuration example in which the prediction device 10 and the terminal 40 are different devices is adopted, but the present invention is not limited to this, and a configuration example in which the prediction device 10 and the terminal 40 are the same device is adopted. good too.
Furthermore, in the present embodiment, a configuration example in which the prediction device 10, the terminal 30, and the terminal 40 are different devices is adopted, but the present invention is not limited to this, and the prediction device 10, the terminal 30, and the terminal 40 are the same device. Configuration examples may be employed.

In addition, if the network 90 used for the connection between the prediction device 10 and the terminal 20, the connection between the prediction device 10 and the terminal 30, and the connection between the prediction device 10 and the terminal 40 is capable of transmitting and receiving data, the type is not particularly limited, and may be, for example, the Internet, LAN (Local Area Network), WAN (Wide Area Network), or the like. A communication line used for data communication may be wired or wireless. Further, the configuration may be such that each device is connected via a plurality of networks 90 or communication lines.

<Hardware configuration example>
FIG. 2 is a diagram showing a hardware configuration example of the prediction device 10 and the terminals 20 to 40. As shown in FIG.
As shown in FIG. 2, each of the prediction device 10, the user's terminal 20, the business operator's terminal 30, and the trader's terminal 40 is a CPU (Central Processing Unit) 100a, which is a calculation means, and a main storage means. and a memory 100c. Each device also includes, as external devices, a magnetic disk device (HDD: Hard Disk Drive) 100g, a network interface 100f, a display mechanism 100d, an audio mechanism 100h, and an input device 100i such as a keyboard and mouse.

The memory 100c and the display mechanism 100d are connected to the CPU 100a via the system controller 100b. Also, the network interface 100f, the magnetic disk device 100g, the audio mechanism 100h and the input device 100i are connected to the system controller 100b via the I/O controller 100e. Each component is connected by various buses such as a system bus and an input/output bus.

A program for realizing each function is stored in the magnetic disk device 100g. Various functions are realized by loading this program into the memory 100c and executing processing based on this program by the CPU 100a.

<Functional configuration of prediction device>
Next, the functional configuration of the prediction device 10 will be described.
FIG. 3 is a diagram showing a functional configuration example of the prediction device 10. As shown in FIG.
The prediction device 10 includes a transmission/reception unit 11 , an attribute information acquisition unit 12 , a lifeline information acquisition unit 13 , a usage suspension information acquisition unit 14 , a prediction unit 15 and an output unit 16 .

The transmitting/receiving unit 11 is a network interface for connecting with the user's terminal 20, the business operator's terminal 30, and the trader's terminal 40 via the network 90 (see FIG. 1). As a result, the prediction device 10 exchanges data with the terminals 20-40.

The attribute information acquisition unit 12 acquires the customer attribute information (customer information) stored in the business operator's terminal 30 (see FIG. 1) via the network 90 (see FIG. 1). to get. The customer attribute information includes information such as the contractor's name, address (address), telephone number, and contractor number.

The lifeline information acquisition unit 13 acquires lifeline information, which is information about a customer's lifeline. The customer's lifeline is intended to include water, gas, electricity, etc., and is synonymous with the infrastructure business.
The lifeline information here refers to information indicating the amount of water, gas, electricity, etc. used or used in a predetermined period, the amount of consumption, and the charge corresponding to the amount of use.

The usage suspension information acquisition unit 14 acquires usage suspension information, which is information about suspension of usage of the customer's lifeline. Suspension of the customer's lifeline includes shutting down water, gas, etc. as well as shutting down electricity.
The use suspension information referred to here includes information indicating that the plug has been closed and information indicating that electricity has been stopped. In addition, the usage suspension information may include information specifying the date when the plug is to be closed.

The prediction unit 15 receives the customer attribute information acquired by the attribute information acquisition unit 12, the lifeline information acquired by the lifeline information acquisition unit 13, and the usage suspension information acquired by the usage suspension information acquisition unit 14 as inputs. A generated machine learning model 50, which will be described later, is used. The machine learning model 50 defines the relationship between lifeline information of other customers and suspension of use of other customers' lifelines, and is updated by verification using prediction results.
The prediction unit 15 predicts suspension of use of other customers' lifelines based on the acquired lifeline information of the other customers using the generated machine learning model 50 . Such predictions are made before other customers make requests to close the plugs. It should be noted that, as a form of request for closing, there are a case of telephone reception and a case of Web reception. Such closure refers to one aspect of suspension of use, and may be suspension of other aspects, such as suspension of use of electric power or water supply.

The output unit 16 outputs information indicating suspension of use of another customer's lifeline predicted using the customer's lifeline information. The customer's residence may become unoccupied due to, for example, moving, and therefore, the information indicating suspension of use indicates that the residence is likely to become vacant. This is the information shown. The output destination is the trader's terminal 40, and is performed via the network 90 (see FIG. 1).

Therefore, with the information output by the output unit 16, it is possible to approach customers who are considering selling their homes before closing. Therefore, by using the information of the output unit 16, it is possible to carry out real estate purchase sales to the owner before the property for sale becomes actualized.
The concept of real estate purchase business, etc. here includes real estate purchase business and real estate brokerage business. This includes the case of real estate brokerage where a real estate agent finds

FIG. 4 is a diagram explaining the machine learning model 50. As shown in FIG. The machine learning model 50 is generated by the prediction unit 15 (see FIG. 3).
The machine learning model 50 shown in FIG. 4 constructs a model for predicting customers who are likely to apply for closure in the near future by performing machine learning using past closure acceptance completion information as teacher data.

In the machine learning model 50 , the input data are attribute information 5 a , lifeline information 5 b , and use suspension information 5 c , which are input to an input layer 51 .
The past lifeline information of the customer who acquired the information on suspension of use is used as training data.
Note that the attribute information 5 a is information acquired by the attribute information acquisition unit 12 . The lifeline information 5b is information acquired by the lifeline information acquisition unit 13, and the usage suspension information 5c is information acquired by the usage suspension information acquisition unit 14. FIG.

In the intermediate layer 52, the attribute information 5a, the lifeline information 5b, and the usage suspension information 5c are combined and associated with each other by arithmetic processing.
In the intermediate layer 52, machine learning is performed using a past part of the attribute information 5a, the lifeline information 5b, and the usage suspension information 5c input to the input layer 51, and verification is performed using the remaining past information. , weighting adjustment.
In the case of machine learning, the date when the plugging is completed is assumed to be the date when the moving is completed. Also, whether or not the cap was closed is indicated by 0/1 as correct data. There is no case where the cap is not closed on the closing date, but correct data is created because machine learning cannot be performed without 0/1 data indicating whether or not the cap was closed. In addition, there is another timing estimate.

The output layer 53 outputs an output solution predicting suspension of use of the customer's lifeline or suspension of use. In this way, suspension of use of the lifeline is predicted from the lifeline information of the customer who is not blocked.

Here, machine learning using the past lifeline information of the customer who acquired the information on suspension of use as training data will be described.
The following is an example of machine learning using the information about the customer's phone when the customer's usage suspension information is acquired as training data. Examples of such data include, among a large number of data, those that have been confirmed to be significant, and are not limited to this.

For example, when the information on the suspension of use of the customer's lifeline is acquired, the prediction unit 15 (see FIG. 3) performs machine learning on the information on the customer's telephone as training data.
More specifically, the information about the customer's lifeline includes information indicating the medium type of contact history with the customer. The types of media referred to here include fixed telephones, mobile telephones, e-mails, SMS (short message service), websites, and the like.

Also, for example, the information about the customer's lifeline includes information indicating whether the telephone number is a fixed telephone or not. More specifically, using a combination of the information that the landline phone is an older property and that a certain amount of time has passed since it was unplugged increases the possibility of plugging.
Also, for example, the information about the customer's telephone includes information indicating that the customer's telephone is a mobile phone. More specifically, the customer's telephone information includes information indicating that the first three digits of the telephone number are 090, 080, etc., indicating that the customer is a mobile phone. It is information that it is easier to conduct sales by telephone and lead to real estate transactions.

For example, information about a customer's lifeline includes information indicating whether or not a telephone number is registered as the customer's attribute information. More specifically, it is information that among customers who have actually closed their accounts, customers who do not have a telephone number are highly likely to be longtime subscribers, and such customers are highly likely to be closed.

In addition, the information on the customer's lifeline includes, as a customer contact history, information indicating when, who, for what purpose, in what medium, and what applied. More specifically, it is information focused on the possibility that a name change or the like may prove that the owner has changed, and that there may be an application from someone other than the contractor.
In addition, the customer's lifeline information includes information indicating that there was no communication associated with the use of the lifeline within a predetermined period. More specifically, it is information that there is a high possibility that a person who does not have a telephone number will not be using the telephone number if the telephone number is actually closed.

Also, the information on the lifeline includes information indicating the elapsed time from the start of use. In other words, it is information as to how many months have passed since the bottle was opened.

In addition, in the lifeline information, the ratio of the usage amount in a predetermined usage period to the usage amount in a predetermined time period in a day is compared with the average value of multiple customers for each customer. It should contain information indicating the result. For example, it is information that the probability of plugging is high when the average value of power data is high in the ratio of the time period in the early morning to the whole. This is the case for customers who tend to be active early in the morning. Since such information changes with changes in lifestyle patterns over time, it is possible to deal with such changes by updating the machine learning model 50 .

In addition, the information on the lifeline includes information indicating that there is a change in the usage amount during a predetermined usage period and the latest usage amount. Such information includes information indicating an increase and information indicating a decrease. By comparing the amount of usage one month ago, three months ago, or the same month one year ago, if it is decreasing, it is information that there is a high possibility that the plug will be closed. If the period during which the plugging is more likely to occur changes on a monthly basis, it can be handled by monthly updating.

A high percentage of customers who have actually stopped using lifelines have already stopped using them. In other words, there is a high possibility that customers who have stopped using the lifeline will actually shut it down. In addition, training data can be considered from the viewpoint of whether or not relatively large-sized products and facilities are installed when using lifelines.

Figure 5 (a) ~ (c) shows the same customer's consumption, (a) is an example of gas consumption in March according to the latest meter reading, (b) is last year's monthly consumption, (c) shows another example of gas consumption in March according to the latest meter reading. For reference, for March (a) and (c), the gas consumption in March (b) is indicated by a dashed line.

Figures 5(a)-(c) show the monthly gas consumption of customers using gas heating. For this reason, there is a seasonal variation in which gas consumption is relatively high from around December to March when it gets cold, and gas consumption is low at other times when gas heating is not used. In addition, the amount of gas consumed to make hot water for daily use also fluctuates seasonally, with the amount of gas consumed increasing during colder months. Therefore, the annual gas consumption fluctuates as shown in FIG. On the other hand, although the amount of gas consumed fluctuates due to whether it is colder than usual or not, the trend of annual gas consumption remains unchanged.

However, if gas consumption is lower than usual or there is no gas consumption, factors other than seasonal fluctuations are assumed. Therefore, by focusing on the change in gas consumption from the previous year, it can be estimated that the move has been completed or that there is a plan to move in the near future.

A specific description will be given below.
In January and February of this year shown in FIG. 5(a), no significant change can be confirmed compared to January and February of last year shown in FIG. 5(b). On the other hand, the gas consumption in March of this year decreased compared to March of last year shown in (b).
Further, no significant changes can be confirmed in January and February of this year shown in FIG. 5(c) compared to January and February of last year shown in FIG. 5(b). On the other hand, there is no gas consumption in March this year, and gas is not consumed.

In this way, changes in the gas usage trend can be seen in the gas consumption shown as March in (a) and (c) of the figure. Extraction of customers with changes in gas consumption indicates cases in which gas consumption has decreased compared to the corresponding period in the past, and cases in which there was gas consumption in the corresponding period in the past but this year's gas consumption has increased. It is done even if it is not.

In this way, by comparing the gas consumption with the corresponding period in the past, even if there is a period of absence such as a long business trip in a year, it is possible to cope with the decrease due to seasonal factors, etc., and estimate the vacant house. It is possible to prevent a situation in which the accuracy of is lowered. The case shown in FIG. 5 is applicable to customers whose contract period is one year or more.

Here, a case different from the case of FIG. 5 will be described.
In other words, if the gas heater is not used during the cold season or if the bathroom is not used much throughout the year, the gas consumption is almost constant throughout the year. Therefore, if the gas consumption was normal until last month, but the gas consumption for this month according to the most recent meter reading has decreased compared to the previous month, or if there is no gas consumption for this month, the gas consumption will change. You may make it extract as a customer who had the change of .
In addition, if gas consumption is lower than the previous year during a period in which there are relatively many moves, such as from March to May, the customer will be extracted as a customer with a change in gas consumption. may

Although FIG. 5 describes the case of gas consumption, the present invention is not limited to this, and can be applied to power consumption, water consumption, and the like.

Next, another example of teacher data will be described with reference to FIG.
FIG. 6 is a diagram illustrating another example of teacher data.
As such training data, for example, information such as the age of the building, the distinction between web/telephone access, contract period, etc., or a combination thereof is used. Includes those that take into consideration real estate transaction business that uses prediction results for suspension of lifeline use.

The building age referred to here is information about the building in which the lifeline is used, and is included in the lifeline information. The information about the building includes, for example, information indicating a room, which is a unit for sale in a second-hand condominium, and is a candidate for a real estate transaction sales destination.

Building age is narrowed down to properties that have been built for a certain number of years, and properties that have just been completed are excluded. Newly built properties, which are only a few years old, have not lost much of their asset value compared to newly built properties, and there are not many properties available for sale. On the other hand, for second-hand properties that have passed a certain number of years, it is possible to further increase the asset value by renovating them to standard specifications. In addition, it is possible to reflect, to some extent, whether or not a building meets the new seismic standards, and the characteristics of a building that does not use asbestos, depending on the age of the building. In addition, since the new earthquake resistance standards apply to buildings whose construction confirmation date is after June 1, 1981, the upper limit of building age is set from the viewpoint of whether or not the new earthquake resistance standards are applied. be able to.

The difference between web closure/telephone closure is a method of requesting closure, and by distinguishing between web closure and telephone closure, added value as sales materials can be increased. To give an example, in the case of web blocking, it is estimated that the proportion of young people is generally relatively high, and in the case of telephone blocking, it is estimated that the proportion of elderly people is generally relatively high. .
In the case of web closures, it is estimated that the percentage of people who are good at collecting information on the Internet is high. It is believed that there is. In addition, it is considered that there is a high business affinity when conducting telephone sales of real estate transactions to customers who are closed by telephone.
We have explained the distinction between turning off the gas on the web and turning off the telephone, but there are both the web and the telephone for turning off the use of electricity.
To explain further, in some cases requests for turning on gas and starting to use electricity can be made not only by telephone but also on the Web. For this reason, there is a certain percentage of customers who request both when to start using energy and when to stop using energy. can find out.

Including the contract period as a condition is that in the case of rentals, there is a tendency to move out and cancel the energy contract at the time of contract renewal or shortly before the contract renewal. is likely to be shorter than for residents of condominiums, so by excluding customers with short-term energy contracts, the percentage of projects with actual demand will be increased.
The contract period referred to here may include, in addition to the contract for gas, a contract for other than gas, such as electricity. Due to the liberalization of electric power, even gas companies have attribute information of customers who have contracts for both gas and electric power or only for electric power.

A specific description will be given with reference to FIG.
(1) and (2) in FIG. 6 combine building age and contract period. That is, in the case of (1), it is "15 years old or more and the contract period is 3 years or more", and in the case of (2), it is "10 years or more and the contract period is 3 years or more". As a result, it is possible to create a group that is highly likely to include condominiums that are relatively old and that can further increase the asset value of the property.
Note that the contract period in (1) to (2) above is the contract period for gas, and is different from, for example, (10) to (11) described later, which also includes the contract period for electric power.

In addition, for condominiums, there are so-called cases of actual demand where the purchaser or related parties reside, and so-called investment cases where the purchaser or related parties do not live in and rent out, but it is difficult to narrow down to actual demand. be.

In (3) to (6) of FIG. 6, the building age, contract period, and closing request method are combined. In other words, there are four combinations of "contract period of 3 years or more", building age of "5 years or more" and "15 years or more", and closing request method of "web closing" and "telephone closing". (3) to (6) above.

To explain further, in the case of (3) to (4) above "15 years or more and contract period of 3 years or more", in addition to the case of residents who have been living since the time of new construction, due to subsequent sales etc. The case of a new resident whose resident has changed may also be included.
On the other hand, in the case of (5) to (6) above, “5 years or more and contract period of 3 years or more”, compared to (3) to (4) above, the higher percentage of residents living in In addition, assuming that condominiums tend to be sold within 10 years rather than over 10 years old, in the case of (5) to (6) above, the building must be at least 5 years old. Therefore, it can be expected that the percentage of customers who are thinking of selling their current condominium for sale will be higher than the above (3) to (4) by combining with the information on the closing request.

In (7) to (8) of Fig. 6, "five years or more in the building and four years or more in the contract period" are separated when the closing request method is "web closing" and "telephone closing". . Since the difference between the building age and the contract period is smaller than in the above cases (5) to (6), the percentage of residents who have moved in from the time of the new construction is even higher. In general, the number of times people experience buying and selling real estate in their lives is not many. ) to (6).

(9) in FIG. 6 is "15 years or more in building year and 5 years or more in contract period" in the case of "web closure", and the threshold for contract period is two years longer than in (3) above. As a result, residents of rental apartments are excluded more than in the case of (3) above, and it is expected that the ratio of actual demand projects will be increased. In addition, since the above-mentioned (9) requires "Web closure" as a condition, it is conceivable that the effect of Internet advertisement is high as described above.

(10) to (11) in FIG. 6 show customers with a contract period of 3 years or more and a gas contract and an electric power contract among the customers who are ``five years old or more and have a contract period of 3 years or more''. '' and ``telephone blocking'' are separated.
The above (10) to (11) are cases of contractors who are not only gas contracts but also electric power contracts due to the liberalization of electric power. In general, customers who have contracts for gas and electricity with gas suppliers have a higher level of trust in gas suppliers than customers who have contracts for only one of them. It can be expected that a high percentage of people will be interested in listening to the explanation until the end when conducting telephone sales of real estate transactions using the name of the gas company.

To explain further, since the building age and contract period of (10) to (11) above are the same as (5) to (6) above, the above (5) and (10) or the above By combining (6) and (11), it is possible to list customers who only have gas contracts, customers who have both gas and electric power contracts, and customers who only have electric power contracts. In the case of a customer who only has an electricity contract, it is highly likely that the property is an all-electric property.
Also, if you have a power contract, the gas is turned off, but if you have a power contract, you may still return to your home regularly, but it is likely that the house will become vacant in the future. . If the contract for both electricity and gas ends, it is highly likely that the apartment is currently presumed to be vacant.

(12) to (13) in Fig. 6 show that "2 years or more in the building year and 2 years or more in the contract period" are different depending on whether the closing request method is "web closing" or "telephone closing". . Since the year of construction and the contract period are the same, the percentage of residents who have moved in from the time of new construction is even higher than in (7) to (8) above.

(14) and (15) in FIG. 6 show "power contractors with a contract period of less than 2 years" separately for "Web closing" and "Telephone closing". The above (14) to (15) differ from the above (12) to (13), which are gas contractors, in that they do not include the age of the building and that they are electric power contractors, so they have different viewpoints. can be a condition.

(16) to (17) in Fig. 6 show that "12 years old or more and contract period of 3 years or more" is set separately for "web closing" and "telephone closing". . The number of cases is larger than in the above cases (3) to (4), which are 15 years or older.

Although illustration is omitted, other information included in the lifeline information will be described.
Such other information includes, for example, various sales activities, periodic inspections (electrical equipment inspections, regular gas safety inspections), equipment repairs, etc. An example of the period after the most recent contact with the resident in the past) can be given.
In addition, as other information, the period from the last installation to the present (installation An example of the period after the time when it was set) can be given.
In addition, regarding the area where the building is located, for example, statistical information on demographic changes such as the outflow rate from the national census conducted to grasp the actual situation of people and households in Japan, the number of each age group, the ratio of household composition, etc. (an example of statistical information about people and/or households). Note that such statistical information may focus on variations in the same area by comparing statistics from a plurality of survey periods, or may focus on differences from comparison of areas in statistics from a single survey period.

In addition, there are the following that focus on sales of real estate transactions.
That is, if the condominium list includes information indicating room size and/or floor plan, it may be determined whether or not the property is highly marketable based on the room size (m ² ) and floor plan information. . For example, it is conceivable to give a large weight to 50 m ² to 80 m ² , which is popular as a practical demand, and to give a small weight to studio rooms of 30 m ² or less.
In addition, if the list of condominiums includes information indicating the nearest station and information indicating the distance from the nearest station directly or indirectly, such as address information such as walking distance information, the nearest station name and walking distance information The weighting may be adjusted based on whether or not the property has high marketability. For example, since Yokohama station has multiple lines and is popular as the nearest station, it is conceivable to increase the weighting. In addition, for example, a 25-minute walk from the nearest station will be traveled using a bus (bus area), and it is assumed that circulation is low, so it is conceivable to reduce the weighting.
Furthermore, if the condominium list contains information on past real estate transactions, past real estate transaction data (including external information) shows that the profit margin per transaction is large from the contract price (or unit price per tsubo). Properties with high operating efficiency may be identified.

In addition, customer attribute information includes contract information with energy companies and other service contract information that can be confirmed (an example of information indicating multiple types of contracts for infrastructure projects) as conditions related to contracts with contractors. In this case, weighting may be performed based on the type and number of contract information. For example, customers who have (or had) contracts for electricity in addition to gas are likely to have high sales efficiency because they have high loyalty to the energy company, so the weight is increased. can be considered. In addition, it is conceivable to do the same when contracting with ancillary services developed by an energy company.

Furthermore, if the customer attribute information includes information indicating the relationship between the energy contractor and the applicant for the suspension of energy use, such as when the relationship between the contractor and the applicant is different, especially if the surname is the same and the relationship is child, etc. In this case, the weight may be increased because there is a high possibility that it is an inherited building. In other words, when a request for gas shutoff is received from a person other than the contractor, if the person has a family relationship such as a relative of the contractor, it is weighted as a possibility of an inherited case. can be considered to be larger. Relatives here refer to blood relatives within the sixth degree, spouses, and relatives by affinity within the third degree.

Next, an application example using the prediction result by the prediction unit 15 will be described.
Such an application example is a case where the output unit 16 (see FIG. 3) makes a determination regarding guidance based on the prediction result. Such guidance is the presence/absence of guidance on whether or not to provide guidance on services that can be provided when the house becomes vacant, and/or the degree of guidance when guidance is provided.

Decisions regarding guidance can be made using the probabilities of the prediction results. For example, if the probability of the prediction result is high, the degree of guidance should be a detailed explanation, and if the probability of the prediction result is low, the degree of guidance should be explained simply. This is a simple explanation.
In addition, it may be determined whether or not to guide. In this case, when it is determined that guidance should be provided, the degree of guidance described above is determined.

Guidance in the application example is provided when a customer applies for closure. There are two types of unblocking applications: by telephone (telephone unblocking) and by the Web (web unblocking). In the following, an application example 1 will be described for determination by telephone blocking, and an application example 2 for determination by Web blocking will be described.
In addition, as for the timing of judgment, there are cases where judgment is made in advance and responds based on the preliminary judgment results when there is an application, and when there is an application, judgment is made about the customer and based on the judgment results Although there are corresponding cases and cases, application examples will be described without distinguishing between them.

FIG. 7 is a diagram showing an example of a screen output and displayed on the business operator's terminal 30 (see FIG. 1) in the case of telephone reception. (b) partially shows another example of the reception screen, and (c) partially shows another example of the reception screen. That is, (b) and (c) show only the region 62 of (a). Note that the closing request in FIG. 7 is used in the sense of stopping the use of gas, but is not limited to this, and the same applies to other infrastructures such as stopping the use of electricity.
The reception screen 60 shown in FIG. 7A has an area 61 for displaying customer information and an area 62 positioned below the area 61 for displaying information on guidance.

In an area 61 of the reception screen 60, in order from the top, "customer information requesting closure" is displayed, an input area 61a for entering the contractor number, and the contractor number input in the input area 61a are associated with each other. and an area for displaying the "name" and "address" of the user. The contractor number or name referred to here is an example of specific information that identifies the contractor, as described above.
In the example shown in FIG. 7(a), the telephone operator who is answering the phone has input the contractor number "○○○○○" heard from the customer on the telephone T30 (see FIG. 1) into the input area 61a. . Then, in the area 61, the "name" corresponding to the contractor number "○○○○○" is "□□□□□", and the "address" is "△△△ prefecture △△△ city △△ ward △ △1-1-1” is displayed.

The "name" and "address" displayed in the area 61 are for the telephone operator to verify the identity of the customer, and the telephone operator can confirm whether the name and address provided by the customer match those displayed. Verify your identity.

In an area 62 of the reception screen 60, "Guidance on services that can be provided due to becoming a vacant house" is displayed in order from the top, and the determination result for the customer with the contractor number "○○○○○" in the area 61 is displayed. There are regions 63 and . The determination result area 63 includes an indication 63a indicating whether or not guidance is to be provided, and an indication 63b indicating the degree of guidance when guidance is provided.

In FIG. 7A, the display 63a of the region 63 indicating the determination result is "presence/absence of guidance: present", and the display 63b is "level of guidance: detailed". Therefore, the telephone operator refers to the display 63a, 63b in the region 63 showing the determination result after confirming the identity of the customer, and thoroughly explains to the customer on the phone the services that can be provided as the house becomes vacant.
For example, after explaining that we provide real estate purchase or brokerage services, we will ask if you are thinking about selling real estate or if you are interested in this service. Explain the customer benefits of the service and answer customer questions. In addition, at the customer's request, materials such as pamphlets can be sent to the customer's home, or an appointment can be made to have a sales representative visit the customer's home.

In FIG. 7B, the display 63a of the region 63 indicating the determination result is "presence/absence of guidance: present", and the display 63b is "level of guidance: simple". The telephone operator briefly explains to the customer on the phone the services that can be provided as the house becomes vacant.
For example, we will explain that our company provides real estate purchase or brokerage services, and will only tell you that we would like you to contact us if you are interested.

In FIG. 7(c), the display 63a of the region 63 indicating the determination result is "presence/absence of guidance: no", and the display 63b is "-". The telephone operator does not explain to the customer who is on the phone the service that can be provided as the house becomes vacant, and only accepts the closing request.

In this way, since the presence or absence of guidance and the degree of guidance are displayed in the area 63 indicating the judgment result, the telephone operator responds to the telephone according to the display of the area 63, and makes a call requesting closing. It is possible to reliably guide real estate transactions to those customers who are determined to be advised of real estate transactions.
In addition, when providing guidance, for example, by providing guidance according to the degree of guidance determined according to the estimation result of the possibility of a vacant house, it is possible to provide a firm explanation to customers with a high possibility of vacant houses, Real estate transactions can be conducted efficiently by simply explaining to customers with low potential.

8 and 9 are diagrams showing examples of screens output and displayed on the user's terminal 20 (see FIG. 1) in the case of Web reception. FIG. 8(a) shows the Web reception screen (1 /n). In addition, FIG. 8(b) shows an example of a Web reception screen (2/n) for closing application, FIG. 9(a) shows another example of the reception screen (2/n), and (b) , shows another example of the reception screen (2/n). "n" here is an integer of 2 or more.
The reception screen 70 shown in FIG. 8(a) has a message to the customer, "I accept the closing application" and "Please enter the contractor number", and an area 71 for displaying customer information.

An area 71 of the reception screen 70 has an input area 71a for inputting the contractor number, and an area for displaying the "name" and "address" associated with the contractor number input to the input area 71a. be. In the example shown in FIG. 8(a), the same "name" and "address" are displayed as in the case shown in FIG. 7(a).
The user confirms the contents and, if correct, operates the OK button. Thereby, it progresses to the screen of FIG.8(b).

For security purposes, a predetermined personal identification number or password may be entered, or the screen shown in FIG. 8A may be displayed after the login operation. A personal identification number and information to be input when logging in are examples of specific information that identifies a contractor.

The reception screen 70 shown in FIG. 8(b) has an area 72 for the desired closing date and an explanation display area 73 for displaying an explanation of the real estate transaction service.
The area 72 has, in order from the top, a display of "Please input the desired closing date" and an area 72a for the user to input the desired closing date. In the example shown in FIG. 8(b), the area 72a is an unfilled blank.

In the explanation display area 73, there are, in order from the top, a display of "guidance on real estate transaction services" and an area 73a surrounded by a dashed-dotted line for indicating the display range of the guidance. To explain further, in the example shown in FIG. 8(b), the determination result for the user is "presence/absence of guidance: yes" and "extent of guidance: detailed", and the text of the guidance is long. For this reason, a scroll bar 73b is displayed in the area 73a indicating the contents of the guidance to display the part that is not displayed by shifting the notation in the explanation display area 73 up and down.

The reception screen 70 shown in FIG. 9(a) has an area 72 for the desired closing date and an explanation display area 73 for displaying an explanation of the real estate transaction service. The area 72 is the same as in the case shown in FIG. 8(b), and its explanation is omitted.
As in the case shown in FIG. 8(b), the explanation display area 73 has, in order from the top, the display of "real estate transaction service information" and an area 73a. Another example shown in FIG. 9(a) is a case where the determination result for the user is "presence/absence of guidance: yes" and "level of guidance: simple", and the text of the guidance is shown in FIG. 8(b). short in comparison. Therefore, the scroll bar 73b shown in FIG. 8B is not displayed in the area 73a indicating the content of guidance.

The reception screen 70 shown in FIG. 9(b) has an area 72 for the desired closing date. The area 72 is the same as in the case shown in FIG. 8(b), and its explanation is omitted.
To explain further, another example shown in FIG. 9B is a case where the determination result for the user is "presence/absence of guidance: no" and "extent of guidance: -", and the notation of guidance is omitted. be. Therefore, the reception screen 70 shown in FIG. 9(b) does not have the explanation display area 73 shown in FIGS. 8(b) and 9(a). In the explanation display area 73, the length of the explanation is made different (the length of the explanation).

In this way, real estate transaction guidance can be reliably provided in the area 73a of the explanation display area 73 to those customers who are determined to be advised of the real estate transaction among the customers of the Web request. Further, when providing guidance, the content of the guidance is changed according to the degree of guidance. In other words, a simple explanation is given assuming that the customer who is unlikely to become an empty house due to moving is likely to have no plans to move, even though the information is provided to the customer. On the other hand, for customers who are likely to become vacant due to moving, they are highly likely to be interested in guidance on real estate transactions, so detailed explanations are provided while summarizing the main points. As a result, efficient real estate transaction guidance can be provided on the screen for making a Web request. Note that attention is focused on the possibility of a house becoming vacant due to moving, but the degree of guidance may also be determined according to the customer's age group, along with or instead of it.
By omitting real estate transaction guidance for customers who are unlikely to move, it is possible to reduce the burden of screen confirmation associated with customer operations.

Although the embodiments of the present invention have been described above, the technical scope of the present invention is not limited to the scope described in the above embodiments. It is clear from the scope of claims that various modifications or improvements to the above embodiment are also included in the technical scope of the present invention.

DESCRIPTION OF SYMBOLS 10... Prediction apparatus 12... Customer information acquisition part 13... Lifeline information acquisition part 14... Use suspension information acquisition part 15... Prediction part

Claims (10)

an attribute information acquisition unit that acquires customer attribute information;
a lifeline information acquisition unit that acquires information about a customer's lifeline;
a suspension of use information acquisition unit that acquires information about the suspension of use of the customer's lifeline;
By machine learning the attribute of the customer who acquired the information on the suspension of use and the information on the past use of the lifeline as training data, the other customer's lifeline from the acquired attributes and information on the lifeline a prediction unit that predicts suspension of use of
prediction device with causing the prediction unit to machine-learn information about the customer's phone call as the teacher data when information about the suspension of use of the customer's lifeline is acquired;
A prediction device according to claim 1 . The information on the customer's lifeline includes information indicating the medium type of contact history with the customer,
A prediction device according to claim 2 . The information about the customer's lifeline includes information indicating whether the telephone number is a fixed telephone or not.
A prediction device according to claim 2 . The information on the customer's lifeline includes information indicating whether or not a telephone number is registered as the customer's attribute information.
A prediction device according to claim 1 . The information on the customer's lifeline includes information indicating when, who, for what purpose, what medium, and what was applied as a customer contact history,
A prediction device according to claim 1 . The information on the lifeline includes information indicating the elapsed time from the start of use,
A prediction device according to claim 1 . In the lifeline information, the result of comparing each customer with the average value of a plurality of customers regarding the ratio of the usage amount in a predetermined usage period to the usage amount in a predetermined time period in a day contains information indicating the
A prediction device according to claim 1 . The information on the lifeline includes information indicating that there is a change in the usage amount in a predetermined usage period and the trend of the most recent usage amount.
A prediction device according to claim 1 . information processing equipment,
an attribute information acquisition function for acquiring customer attribute information;
a lifeline information acquisition function for acquiring information about a customer's lifeline;
a suspension of use information acquisition function for acquiring information regarding the suspension of use of the customer's lifeline;
By machine learning the attribute of the customer who acquired the information on the suspension of use and the information on the past use of the lifeline as training data, the other customer's lifeline from the acquired attributes and information on the lifeline A prediction function that predicts the suspension of use of
program to realize

Images