Estimation model for estimating an attribute of an unknown customer

Patent metadata

Field	Value
Publication number	US-10997612-B2
Application number	US-201514974565-A
Country	US
Kind code	B2
Filing date	Dec 18, 2015
Priority date	Dec 19, 2014
Publication date	May 4, 2021
Grant date	May 4, 2021

Abstract

There is provided a generating apparatus that generates an estimation model for estimating an attribute of an unknown target, the generating apparatus including a training-data acquiring unit acquiring training data including an attribute of a known target, attributes of other targets related to the known target, and a plurality of sets of types of relations between the known target and the other targets and a training processing unit training, using the training data, on the basis of the attribute of the known target and a type of a relation between the known target and the unknown target related to the known target, an estimation model for estimating the attribute of the unknown target, a generating method using the generating apparatus, and a program used in the generating apparatus.

First claim

What is claimed is:

1. An apparatus that generates an estimation model for estimating an attribute of an unknown customer that is related to a known customer, the apparatus comprising:
at least one hardware processor accessing memory to produce a trained estimation model by
a) a training-data acquiring unit acquiring a first set of training data including an attribute of a known customer, attributes of other customers related to the known customer, and a plurality of sets of types of relations between the known customer and the other customers and a second set of training data including at least one segment the known customer is classified, wherein the segment is an indication of the known customer's purchase desire;
b) a training processing unit training an estimation model for estimating the attribute of the unknown customer, using the first set of training data and the second set of training data, on a basis of the attribute of the known customer and a type of a relation between the known customer and the unknown customer and attributes of a sales representative and relation state information between the sales representative and the known customer;
c) a segmentation-condition storing unit storing, in association with each of at least one segment, a plurality of segmentation conditions that are true in response to classifying the known customer into the at least one segment based on identifying at least one attribute of the unknown customer;
d) a model generating unit generating a state transition model as a regression tree based on a plurality of the known customers and an optimum policy in the state transition model to maximize a reward from the unknown customer and supplying the second set of training data to the training-data acquiring unit;

the at least one hardware processor accessing memory to execute the state transition model by
e) an estimating unit estimating, using the state transition model, the trained estimation model, and the optimal policy, the at least one segment of the unknown customer on a basis of the attribute of the known customer and a type of a relation between the known customer and the unknown customer to calculate a probability that the unknown customer belongs to each of a plurality of segments, and updating the probability that the unknown customer belongs to each of the plurality of segments after a type of sales communication is performed on the unknown customer and repeating the estimating by the estimation unit until the reward is above a settable reward threshold or the probability that the unknown customer belongs to one of the plurality of segments is above a settable probability threshold;
f) a determining unit determining, based on the trained estimation model, and the optimum policy and according to a probability that the unknown customer belongs to each of the plurality of segments, a type of sales communication to be performed on the unknown customer by the sales representative, wherein the type of sales communication to be performed on the unknown customer by the sales representative is one of face-to-face sales activities, sales activities by telephone, and commodity introduction by mail; and
g) an output unit outputting a processing result of the at least one segment of the unknown customer obtained from the estimating unit and the type of sales communication on the unknown customer obtained from the determining unit.

2. The apparatus according to claim 1 , wherein
the training-data acquiring unit acquires, as the attributes of the other known customers, segments into which the other known customers should be classified among a plurality of segments, and
the training processing unit trains the estimation model for estimating a segment into which the unknown customer should be classified.

3. The apparatus according to claim 2 , wherein the training-data acquiring unit acquires a profile of the known customer as the attributes of the known customer.

4. The apparatus according to claim 1 , wherein
the training-data acquiring unit acquires the first set of training data further including an attribute of the sales representative that performs the type of sales communication, and
the training processing unit trains the estimation model further on a basis of the attribute of the sales representative.

5. The apparatus according to claim 4 , wherein
the training-data acquiring unit acquires the first set of training data further including relation state information of the known customer with the sales representative that performs the type of sales communication, and
the training processing unit trains the estimation model further on a basis of the relation state information.

6. The apparatus according to claim 1 , further comprising a segmentation-condition storing unit storing, in association with each of at least one segment, a plurality of segmentation conditions that should be satisfied by a profile of a customer classified into the segment, wherein
on condition that a part of profiles of the unknown customer are found and a part of the plurality of segmentation conditions are satisfied, the estimating unit updates the probability that the unknown customer belongs to each of the segments.

7. The apparatus according to claim 4 , wherein
the training-data acquiring unit acquires first set of training data including an attribute of a known customer as the attribute of the known customer, including attributes of other customers related to the known customer as the attributes of the other customers, and including, as the plurality of sets of types of the relations between the known customer and the other known customers, types of relations between the known customer and the other customers disclosed to the action entity by the known customer at a point when the other customers are unknown, and
the training processing unit trains, on a basis of the attribute of the known customer, the attributes of the other customers, and the plurality of sets of types of the relations between the known customer and the other customers, the estimation model for estimating a segment into which the unknown customer having the type of the relation with the known customer should be classified.

8. The apparatus according to claim 1 , wherein the training-data acquiring unit acquires, as the types of the relations between the known customer and the other known customers, the first set of training data including a type of a human relation.

9. A computer-implemented method for generating an estimation model for estimating an attribute of an unknown customer that is related to a known customer, the method comprising the steps of:
producing a trained estimation model by steps a through d
a) acquiring a first set of training data including an attribute of a known customer, attributes of other customers related to the known customer, and a plurality of sets of types of relations between the known customer and the other customers and a second set of training data including at least one segment the known customer is classified, wherein the segment is an indication of the known customer's purchase desire;
b) training an estimation model for estimating the attribute of the unknown customer, using the first set of training data and the second set of training data, on a basis of the attribute of the known customer and a type of a relation between the known customer and the unknown customer and attributes of a sales representative and relation state information between the sales representative and the known customer;
c) storing, in association with each of at least one segment, a plurality of segmentation conditions that are true in response to classifying the known customer into the at least one segment based on identi

Assignees

• IBM

Inventors

• Amano Shunichi
• Kamijoh Kohichi
• Kawahara Ryo
• Ono Masaki

Classifications

• G06Q30/0204

  Market segmentation · CPC title

• G06Q30/0201 Primary

  Market modelling; Market analysis; Collecting market data · CPC title