Policy planning support system and policy planning support method

Abstract

To support consideration of policies even when there is no history of policy implementation.SOLUTION: A policy planning support system is equipped with a processor and a storage device. The storage device holds target information, including multiple feature values for each of a plurality of targets. The multiple feature values include at least one of the feature values related to the behavior of the targets and the feature values related to the attributes of the targets. The processor identifies two or more feature values among the multiple feature values that are related to a feature value specified as an evaluation index as related feature values, projects the related feature values of each target onto a space with each of the two or more related feature values as an axis, selects one or more targets whose evaluation based on the feature value specified as the evaluation index meets a predetermined criterion as role models, and outputs information for displaying the space onto which the related feature values of the plurality of targets, including the role models, are projected as information for supporting policy planning to improve the evaluation index.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a support system and method for providing information necessary for planning measures. [Background technology]

[0002] Measures that influence customers or target individuals (hereafter referred to as "target individuals") with the aim of improving evaluation indicators (Key Performance Indicators, hereafter referred to as "KPIs") are considered important in a wide range of industries for purposes such as promoting marketing, improving learning efficiency, and improving advertising effectiveness. In order to plan useful measures, it is necessary to estimate the effect of the measures and select the target individuals who will actually apply the measures. If measures are implemented without estimating the effect of the measures and the target individuals who can actually apply the measures are selected, the special feeling of the measures may be lost, and the effects of the measures may be diluted.

[0003] Technologies for supporting the implementation of policies include those described in, for example, Patent Publication No. 2016-118975 (Patent Document 1) and Patent Publication No. 2021-26534 (Patent Document 2).

[0004] In Patent Document 1, in order to support the implementation of marketing measures, the effectiveness is estimated by referring to the implementation history of measures that records the effects of measures implemented in the past, such as changes in KPIs, while optimization is performed to maximize sales and profits defined as KPIs.

[0005] In Patent Document 2, a sales promotion support device estimates the effect of a certain measure when it is implemented on similar targets, for a certain measure that has been implemented on at least one target among a plurality of targets.

[0006] Both of these documents assume that the history of past policy implementation, or the results of past policy implementation, is kept in order to estimate the effectiveness and select policy recipients. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-118975 [Patent Document 2] Patent Publication No. 2021-26534 Summary of the Invention [Problem to be solved by the invention]

[0008] The impact on KPIs of implementing measures for targets who are expected to benefit is estimated and considered based on the implementation history of past measures. In Patent Document 1, when optimizing marketing measures, it is necessary to proceed while predicting the effects of the measures based on the implementation history of at least multiple measures.

[0009] Similarly, in Patent Document 2, it is necessary to refer to the implementation history of other subjects in order to provide sales promotion support, and the method involves selecting measures that are expected to be effective from among measures that have been implemented in the past.

[0010] Therefore, it is difficult to apply prior art when formulating policies.

[0011] Therefore, an object of the present invention is to provide information for planning measures that are expected to be effective even when there is no history of implementing such measures. [Means for solving the problem]

[0012] In order to solve at least one of the above problems, the policy planning support system of the present invention comprises a processor and a storage device, wherein the storage device holds subject information including a plurality of feature amounts related to each of a plurality of subjects, the plurality of feature amounts including at least one of feature amounts related to the behavior of the subject and feature amounts related to the attributes of the subject, and the processor identifies, from the plurality of feature amounts, two or more feature amounts that are related to feature amounts specified as evaluation indexes as related feature amounts, projects the related feature amounts of each of the subjects onto a space whose axes are each of the two or more related feature amounts, selects, from the plurality of subjects, one or more subjects whose level of evaluation based on the feature amounts specified as evaluation indexes meets a predetermined standard as role models, and outputs information for displaying the space onto which the related feature amounts of the plurality of subjects, including the role models, are projected, as information to support policy planning for improving the evaluation indexes. [Effects of the Invention]

[0013] According to one aspect of the present invention, feature quantities highly related to KPIs are extracted from data on the characteristics and traits of subjects, and the feature quantities of the subjects are mapped on a space with the feature quantities as axes. Among these, subjects with particularly high KPIs are set as role models, and the system limits the scope to subjects who exist in an area close to the role models, thereby assisting in the consideration of measures to bring the axis items closer to the role models.

[0014] Problems, configurations, and effects other than those described above will become apparent from the following description of the embodiments. [Brief explanation of the drawings]

[0015] [Figure 1] FIG. 2 is a block diagram showing an example of a functional configuration of the policy planning support system. [Figure 2] This is a list of parameters stored in the database. [Figure 3A] FIG. 10 is an explanatory diagram of a feature amount table. [Figure 3B] FIG. 10 is an explanatory diagram of KPI information. [Figure 3C] FIG. 10 is an explanatory diagram of a related feature amount. [Figure 3D] FIG. 10 is an explanatory diagram of related feature amount selection conditions. [Figure 3E] FIG. 10 is an explanatory diagram of target person information. [Figure 4A] FIG. 10 is an explanatory diagram of spatial information. [Figure 4B] FIG. 10 is an explanatory diagram of role model information. [Figure 4C] FIG. 10 is an explanatory diagram of role model selection conditions. [Figure 4D] FIG. 10 is an explanatory diagram of policy recipient information. [Figure 4E] FIG. 10 is an explanatory diagram of policy recipient selection conditions. [Figure 4F] FIG. 10 is an explanatory diagram of integrated spatial information. [Figure 4G] FIG. 1 is an explanatory diagram of integrated spatial conditions. [Figure 5A] FIG. 10 is an explanatory diagram of policy information. [Figure 5B] FIG. 10 is an explanatory diagram of transition information. [Figure 5C] FIG. 10 is an explanatory diagram of effect confirmation index information. [Figure 5D] FIG. 2 is an explanatory diagram of setting information. [Figure 6] FIG. 10 is an explanatory diagram showing an example of visualization by a mapping unit. [Figure 7] FIG. 10 is an explanatory diagram showing an example of confirming the effectiveness of a policy. [Figure 8] 10 is an explanatory diagram showing an example of displaying information necessary for policy planning via a graphical user interface in the visualization unit. FIG. [Figure 9] FIG. 2 is a block diagram illustrating an example of a hardware configuration of the policy planning support system. [Figure 10] 10 is a flowchart showing the overall processing procedure of the policy planning support system. [Figure 11] 10 is a flowchart showing a processing procedure of a KPI selection unit. [Figure 12] 10 is a flowchart showing a processing procedure of a feature amount selection unit. [Figure 13] 10 is a flowchart showing a processing procedure of a mapping unit. [Figure 14] 10 is a flowchart showing a processing procedure of a feature amount projection unit. [Figure 15] 10 is a flowchart showing a processing procedure of a role model selection unit. [Figure 16] 10 is a flowchart showing a processing procedure of a policy recipient selection unit. [Figure 17] 10 is a flowchart showing a processing procedure of a space integration unit. [Figure 18] 10 is a flowchart showing a processing procedure of a feature amount projection unit. [Figure 19] 10 is a flowchart showing a processing procedure of a policy effect confirmation unit. [Figure 20] 10 is a flowchart showing a processing procedure of a visualization unit. [Figure 21A] FIG. 3 is an explanatory diagram of a feature amount table in the first embodiment. [Figure 21B] FIG. 2 is an explanatory diagram of setting information in the first embodiment. [Figure 21C] FIG. 10 is an explanatory diagram of KPI information according to the first embodiment. [Figure 22A] FIG. 2 is an explanatory diagram of subject information in the first embodiment. [Figure 22B] FIG. 2 is an explanatory diagram of a related feature amount in the first embodiment. [Figure 22C] FIG. 10 is an explanatory diagram of related feature quantity selection conditions in the first embodiment. [Figure 22D] FIG. 2 is an explanatory diagram of spatial information in the first embodiment. [Figure 23A] FIG. 10 is an explanatory diagram of role model information in the first embodiment. [Figure 23B] FIG. 1 is an explanatory diagram of role model selection conditions in the first embodiment. [Figure 23C] FIG. 10 is an explanatory diagram of policy recipient information in the first embodiment. [Figure 23D] FIG. 10 is an explanatory diagram of policy recipient selection conditions in the first embodiment. [Figure 23E] FIG. 2 is an explanatory diagram of integrated spatial information in the first embodiment. [Figure 23F] FIG. 1 is an explanatory diagram of an integrated spatial condition in the first embodiment. [Figure 24A] FIG. 2 is an explanatory diagram of a visualization unit in the first embodiment. [Figure 24B] FIG. 10 is an explanatory diagram of policy information in the first embodiment. [Figure 24C] FIG. 10 is an explanatory diagram of a display example of transition information in the first embodiment. [Figure 25A] FIG. 10 is an explanatory diagram of a feature amount table in the second embodiment. [Figure 25B] FIG. 10 is an explanatory diagram of setting information in the second embodiment. [Figure 26] FIG. 10 is an explanatory diagram of selection of policy recipients in the second embodiment. [Figure 27A] FIG. 10 is an explanatory diagram of a display example of transition information in the second embodiment. [Figure 27B] FIG. 10 is an explanatory diagram of editing of a feature amount table in the second embodiment. [Figure 28A] FIG. 11 is an explanatory diagram of a web advertisement in a third embodiment. [Figure 28B] FIG. 11 is an explanatory diagram of a feature amount table in the third embodiment. [Figure 29A] FIG. 11 is an explanatory diagram of setting information in the third embodiment. [Figure 29B] FIG. 10 is an explanatory diagram of selection of policy recipients in the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0016] <Example of functional configuration of policy planning support system> FIG. 1 is a block diagram showing an example of the functional configuration of a policy planning support system.

[0017] The policy planning support system 100 supports the planning of policies aimed at improving KPIs by limiting the targets to which the policies (hereinafter also referred to as policy recipients) are applied to those who are expected to see particular benefits. Specifically, the policy planning support system 100 has a data input / output unit 101, a database 102, a KPI selection unit 103, a feature selection unit 104, a mapping unit 105, a spatial integration unit 109, a policy selection unit 110, a policy effect confirmation unit 111, and a visualization unit 112. The mapping unit 105 has a feature projection unit 106, a role model selection unit 107, and a policy recipient selection unit 108.

[0018] The policy planning support system 100 can access a database 102. The database 102 is stored in the policy planning support system 100 or in an external computer (not shown) that can communicate with the policy planning support system 100.

[0019] The database 102 stores the parameters shown in Fig. 2. Sensor information and logs obtained by an external computer, and information calculated within the policy planning support system 100 are also stored.

[0020] The data input / output unit 101 accepts data input from an input device such as a keyboard or an external computer that can communicate with the policy planning support system 100. The data input / output unit 101 also displays data calculated within the policy planning support system 100 on an output device such as a display.

[0021] The KPI selection unit 103 acquires the set feature amount from the data input / output unit 101 or a setting file, the target for the set feature amount, and the feature amount table 200, and determines whether the set feature amount can be selected as a KPI based on the information stored in the feature amount table 200. If a feature amount selectable as a KPI is set, the set feature amount and the target for the set feature amount (such as increase, decrease, or maintain) are stored in KPI information 201.

[0022] The feature selection unit 104 acquires the selection conditions for related feature quantities set from the data input / output unit 101 or a setting file, the relevance with KPIs, the feature table 200, and the KPI information 201 stored in the KPI selection unit 103, calculates the relevance with KPI information 201 based on the index set by the relevance, selects multiple feature quantities that satisfy the selection conditions for related feature quantities, and stores them in the related feature quantities 202. Examples of the relevance include the correlation between KPIs and feature quantities, and the contribution calculated when a machine learning model or the like infers a KPI from the feature quantities, but the index is not limited as long as it indicates the relationship with the KPI. The feature quantities may be continuous or discrete values.

[0023] The related feature amount 202 is defined by the related feature amount selection condition 203. The related feature amount selection condition 203 is not limited to an index as long as it can determine whether or not a feature amount 202 is related. For example, the related feature amount selection condition 203 may be determined based on the relevance to a KPI, or may be determined based on the upper limit number of related feature amounts 202 to be selected.

[0024] The mapping unit 105 includes a feature projection unit 106, a role model selection unit 107, and a policy recipient selection unit 108.

[0025] The feature projection unit 106 acquires the subject information 204 and the related features 202 stored by the feature selection unit 104, and projects, from the subject information 204, the features of each subject that correspond to the related features 202 onto a space whose axis is the related features 202. When there are multiple related features 202, the feature projection unit 106 projects the related features 202 of each subject onto a single or multiple spaces whose axes are some or all of the combinations of the related features 202.

[0026] The role model selection unit 107 stores in role model information 206, among the subjects projected by the feature projection unit 106 onto a space whose axis is the related feature 202, subjects (hereinafter referred to as role models) who belong to an area where subjects who are particularly highly related to KPIs gather. When there are multiple related feature 202, an area (hereinafter referred to as role model area) where role models corresponding to the combinations gather is defined by role model selection conditions 207. The role model selection conditions 207 are not limited to indicators as long as they can determine whether or not a person is a role model. For example, the role model selection conditions 207 may be determined based on the relevance to KPIs or the upper limit on the number of selected role models. When there are multiple related feature 202 and multiple spaces onto which the related feature 202 of the subjects are projected, the role model selection unit 107 stores role model information 206 for each space.

[0027] The policy recipient selection unit 108 stores in policy recipient information 208 subjects (hereinafter referred to as policy recipients) who belong to an area (hereinafter referred to as policy recipient area) close to the area where role models acquired by the role model selection unit 107 gather. The policy recipient area is defined by policy recipient selection conditions 209. The policy recipient selection conditions 209 are not limited to any index as long as it can determine whether or not a person is a policy recipient. For example, the policy recipient selection conditions 209 may be determined by the similarity (or distance) to the role model area, or by the upper limit number of selected policy recipients.

[0028] Here, we will explain role models and policy recipients. A role model is a subject who has already achieved a sufficiently high KPI, and a policy recipient is a subject whose KPI is not as high as that of the role model but who has related features similar to those of the role model. By implementing measures corresponding to the related features for policy recipients, the related features will become closer to those of the role model, making it easier to achieve the KPI. In other words, the effect of improving KPI by implementing measures corresponding to the related features for policy recipients can be expected to be higher than the effect when implemented for other subjects.

[0029] Role models may also be included among policy adopters, but since role models have already achieved sufficiently high KPIs, it is considered that there is relatively little room for improvement in KPIs through implementing policies, and therefore they do not need to be included among policy adopters.

[0030] Based on the information about the subject in each space acquired by the mapping unit 105, the space integration unit 109 stores, in the integrated space information 210, spatial information to be provided to a system user (i.e., a person who uses the policymaking support system 100 to plan a policy for improving KPIs) as information for policymaking. Whether or not to store spatial information in the integrated space information 210 is defined by the integrated space condition 211. The integrated space condition 211 is not limited to any indicator as long as it can determine whether or not to integrate the spatial information. For example, the integrated space condition 211 may be determined by the similarity of role models in each space, or by the degree of association between the related feature 202 and the KPI information. If there is only one space onto which the related feature 202 of the subject is projected, the space integration unit 109 stores only that one space in the integrated space information 210.

[0031] The policy selection unit 110 acquires the feature table 200, the related features 202 stored in the feature selection unit 104, and the policy recipient information 208 stored in the policy recipient selection unit 108, and acquires policy proposals linked to the related features 202 based on the spatial axes used when determining the policy recipient information 208, and stores them in the policy information 212.

[0032] The policy effect confirmation unit 111 acquires the KPI information 201 stored by the KPI selection unit 103, the related features 202 stored by the feature selection unit 104, and the subject information 204, maps the features of the subject before and after the policy in a space whose axis is the related features 202 used when planning the policy, and stores the subject's spatial movement trajectory and the policy effect in transition information 213.

[0033] The visualization unit 112 can display all of the parameters held by the policy planning support system 100, and can also change the values ​​of the parameters via the data input / output unit 101. The visualization unit 112 can also display the space generated by the mapping unit 105 and the feature amounts of the projected subject.

[0034] FIG. 6 is an explanatory diagram showing an example of visualization by the mapping unit 105.

[0035] The axes 601 and 602 of the space represent the related feature 202, the subject 603 represents the projected subject, the area 604 surrounded by dotted lines represents the role model area, and the area 605 surrounded by dashed lines represents the policy recipient area. Space example 606 is an example of a continuous space, and space example 607 is an example of a space in which the subject's features are discretely expressed.

[0036] In the example of discrete space 607, as in display 610, it is sufficient to simply display how many subjects belong to each square (e.g., a section of a predetermined size), and the position information of the subject 611 projected onto the example of space 607 (e.g., a circle plotted at a position corresponding to the value of the relevant feature of the subject 611, as shown in Figure 6) does not have to be displayed.

[0037] The number of subjects may be displayed by writing numerical values ​​in space as in display 610, or by depicting different colors in space according to numerical values, as in a color map. The same display method as above may be adopted not only for discrete space but also for continuous space.

[0038] The role model area 608 and the policy recipient area 609 in the discrete space are indicated by a dotted line and a dashed-dotted line, respectively. The role model area 608 and the policy recipient area 609 may be set so as to overlap, or may not overlap.

[0039] Example 612 in Figure 6 shows an example of selecting policy recipients when mapping is performed on multiple spaces. Subject 613 shows a subject selected as a role model, and is a diagram showing a case where the role models match in two spaces. That is, in example 612, the three people selected as role models in one space are the same as the three people selected as role models in the other space.

[0040] When the policy recipient selection condition 209 is "similarity of role models in each space," in example 612, the matching rate of subjects belonging to the role model area is high, so axes 614, 615, 616, 617 and policy recipient information 208 are provided to the system user as information about the policy. Note that although the space in Fig. 6 is depicted as a two-dimensional space, it may be a three-dimensional space or a one-dimensional space.

[0041] FIG. 7 is an explanatory diagram showing an example of confirming the effectiveness of a policy.

[0042] Space 701 shows a space in which the relevant features of subjects before the measure are mapped. Space 702 shows a space in which the relevant features of subjects after the measure are mapped. If one of the indicators of the measure effect is the "number of subjects belonging to the role model area," the effect of this measure is 3 people (i.e., the number increased from 3 people before the measure to 6 people after the measure, an increase of 3 people).

[0043] The visualization unit 112 can display all parameters including the role model area and the policy recipient area.

[0044] FIG. 8 is an explanatory diagram showing an example of displaying information required for policy planning via a graphical user interface (GUI) in the visualization unit 112.

[0045] The content described here is merely an example, and the visualization method and operation are not limited to those described here.

[0046] In the example of Fig. 8, a system user operates a mouse pointer 801 to click 802 on a role model area. Information about the person who belongs to the role model area selected by the click 802 is displayed in a pop-out 803.

[0047] The visualization unit 112 can display all of the parameters held by the policy planning support system 100, and can also change the values ​​of the parameters via the data input / output unit 101. Example 804 in FIG. 8 shows an example of displaying the feature table 200 via the visualization unit 112.

[0048] <Database> FIG. 2 is a list of parameters stored in the database 102.

[0049] The database 102 includes a feature table 200, KPI information 201, related features 202, related feature selection conditions 203, subject information 204, space information 205, role model information 206, role model selection conditions 207, policy adopter information 208, policy adopter selection conditions 209, integrated space information 210, integrated space conditions 211, policy information 212, transition information 213, effect confirmation indicator information 214, and setting information 215.

[0050] Next, detailed examples of the parameters in FIG. 2 will be described with reference to FIGS. 3A to 5D.

[0051] FIG. 3A is an explanatory diagram of the feature amount table 200.

[0052] The feature quantity table 200 includes an ID that uniquely identifies a feature quantity, obtainable data items, and information required for supporting policy planning for each item. This information may be set via the data input / output unit 101, or may be read from a configuration file or the like that has been entered in advance. This information may also be a real value, a character string, or a flag that indicates True or False. For example, the feature quantity table 200 manages, as a character string, a flag indicating whether a feature quantity can be selected as a KPI, and a proposed policy for that feature quantity. If there are two or more proposed policies for one feature quantity, multiple proposed policies may be stored. The feature quantity table 200 may be updated via the data input / output unit 101.

[0053] A specific example 300 of the feature table 200 includes a feature ID and variables representing the attributes of the feature. The feature ID is used to specify a specific feature. The feature name, demographic, KPI selection availability, policy availability, and policy proposal represent the attributes of the feature.

[0054] Demographic is a flag indicating whether each feature is demographic information (demographic information, such as age, gender, address, occupation, etc.). KPI selectability is a flag indicating whether each feature can be selected as a KPI. Measure availability is a flag indicating whether a measure can be set for each feature. Measure proposals are proposed measures set for feature proposals where the measure availability is "Yes" (i.e., True). Multiple proposed measures may be set for one feature.

[0055] In the specific example 300 of the feature table 200, the types and number of attributes are not limited to those described above. As long as the feature ID can be associated with each feature, it does not need to be expressed in the form of an integer and may be expressed as a character string such as a hash value. If there are no duplicate feature names, the feature name may be used as the feature ID. The attribute may be a discrete categorical variable, a continuous variable, or a character string. If there is no value for the corresponding attribute, "-" may be treated as a missing value. The missing value is not limited to "Null" or "None," which are synonymous with "-," as long as it is a symbol indicating the absence of a value for the corresponding feature.

[0056] FIG. 3B is an explanatory diagram of the KPI information 201.

[0057] The KPI information 201 includes a KPI ID that uniquely defines the KPI setting, a feature value to be set as the KPI, and a target value for the KPI (e.g., increase, decrease, or maintain). The feature value set for the KPI is changed according to the target value as a measure is implemented. For example, when considering measures for a new product campaign at a supermarket, a feature value possessed by the target user, such as "number of new product purchases," corresponds to a feature value set as a KPI, and "increase" is the target value for the number of new product purchases. An ID is assigned to the combination of the feature value "number of new product purchases" and the target value "increase" as a KPI setting. The feature value and target value selected as the KPI may be set via the data input / output unit 101, or may be read from a setting file or the like. The selectability of the KPI managed in the feature value table 200 is referenced, and if selectable, the set feature value and target value are stored as the KPI information 201. The number of feature values ​​stored in the KPI information 201 is not limited.

[0058] A specific example 301 of the KPI information 201 includes a KPI ID, a feature ID selected for the KPI, and a target for the KPI. The feature ID corresponds to the feature ID stored in the feature table 200. The KPI ID does not need to be expressed in the form of an integer, and may be expressed as a character string such as a hash value, as long as it corresponds to the setting of each KPI. There are no restrictions on how the KPI target is expressed, as long as it is a symbol or character string synonymous with "increase," "decrease," or "maintain." For example, it is sufficient that the KPI target is set to be uniquely determined, such as ">" for "increase," "<" for "decrease," and "=" for "maintain."

[0059] FIG. 3C is an explanatory diagram of the related feature amount 202.

[0060] The related feature 202 includes a related feature ID that uniquely defines the setting of the related feature 202, a KPI registered in the KPI information 201, a feature highly related to the KPI registered in the KPI information 201, and the degree of relatedness between the feature and the KPI. The related feature 202 does not include a feature highly related to the KPI that is stored in the KPI information 201. There is no limit to the number of items that can be selected as long as there is one or more related feature 202.

[0061] A specific example 302 of the related feature 202 includes a related feature ID and the feature ID of a feature that is highly related to a KPI stored as the KPI information 201. The related feature 202 does not include a feature that is highly related to a KPI and is stored in the KPI information 201. The feature ID corresponds to the feature ID stored in the feature table 200. As long as the related feature ID corresponds to the setting of each related feature 202, it does not need to be expressed in the form of an integer and may be expressed as a character string such as a hash value. There is no limit to the number of features stored in the related feature 202.

[0062] FIG. 3D is an explanatory diagram of the related feature selection condition 203.

[0063] The related feature quantity selection condition 203 includes an ID that uniquely defines the condition for selecting related feature quantities registered in the related feature quantity 202, and a selection condition that determines whether a feature quantity is related when selecting the related feature quantity. The related information amount selection condition may be a conditional expression including a fixed real number or a variable real number that changes depending on the situation, or a character string that indicates a specific condition. For example, if the condition for selecting a feature quantity as the related feature quantity 202 is that the degree of relevance between the feature quantity and a KPI registered in the KPI information 201 exceeds a threshold, the related information amount selection condition is "the degree of relevance with the KPI is equal to or greater than a threshold." The related feature quantity selection condition 203 may be set by the data input / output unit 101, or may be read from a configuration file or the like that has been entered in advance. The number of conditions is not specified. Selection conditions may be defined for each KPI, or the same selection conditions may be set for all KPI information.

[0064] A specific example 303 of the related feature selection conditions 203 includes a related feature selection condition ID and a selection condition for each feature. A feature that satisfies the selection condition is selected as the related feature 202. The method and number of conditional expressions used for the selection condition are not limited.

[0065] FIG. 3E is an explanatory diagram of the subject information 204.

[0066] The target person information 204 includes characteristics of each target person. For example, when considering measures for a new product campaign, the target person information 204 may include the date and time of visit, purchased items, and transaction amount obtained in a supermarket. The target person information 204 may also include demographic data such as gender and age.

[0067] A specific example 304 of the subject information 204 includes a subject ID and a feature possessed by each subject. The feature of each subject corresponds to each feature held in the feature table 200. The subject ID does not need to be expressed in the form of an integer, and may be expressed as a character string such as a hash value, as long as it corresponds to each subject's feature. The value of each feature may be a discrete categorical variable or a continuous variable. Furthermore, where no value exists, "-" may be treated as a missing value. Missing values ​​are not limited to "Null" or "None," which are synonymous with "-," as long as they are a symbol indicating the absence of a value for the corresponding feature.

[0068] FIG. 4A is an explanatory diagram of the spatial information 205.

[0069] Spatial information 205 is information that sets up a space onto which the features of each subject are projected in order to select subjects (also referred to as policy recipients) who will implement measures to improve a certain KPI (i.e., bring them closer to the target), and includes an ID that uniquely defines the space setting, information on the KPI that is being sought to bring closer to the target, and information on related features that are set as the axes of the space.

[0070] A specific example 401 of the space information 205 includes a space ID, a KPI ID, and a related feature ID that is highly related to the KPI ID and was set as an axis when the space was created. The feature ID corresponds to the feature ID held in the feature table 200. The KPI ID corresponds to the KPI ID held in the KPI information 201. The space ID does not need to be expressed in the form of an integer, as long as it corresponds to the settings of each space, and may be expressed as a character string such as a hash value.

[0071] FIG. 4B is an explanatory diagram of the role model information 206.

[0072] The role model information 206 includes an ID that uniquely defines the setting of a role model, identification information of the selection conditions used to set the role model among the selection conditions set in the role model selection conditions 207, identification information of the space used to set the role model among the spaces set in the space information 205, and identification information of the subjects selected as role models in the setting of the role model among the subjects set in the subject information 204. Specifically, the role model information 206 includes information on subjects who belong to a role model area that is particularly highly related to the KPI, among the subjects projected onto a space whose axis is the feature indicated by the related feature 202.

[0073] A specific example 402 of the role model information 206 includes a role model ID, a role model condition pattern ID, a space ID, and a subject ID selected as a role model. The role model condition pattern ID corresponds to the role model condition pattern ID held in the role model selection conditions 207. The space ID corresponds to the space ID held in the space information 205. The subject ID corresponds to the subject ID held in the subject information 204. The role model ID does not need to be expressed in the form of an integer, and may be expressed as a character string such as a hash value, as long as it corresponds to each role model, the role model selection conditions 207 when the person was selected as a role model, and the space information 205.

[0074] FIG. 4C is an explanatory diagram of the role model selection conditions 207.

[0075] The role model selection conditions 207 include an ID (role model condition pattern ID) that uniquely defines the setting of the selection conditions, and a selection condition that defines whether or not a subject is to be a role model. The role model selection conditions 207 may be a conditional expression that includes a fixed real number or a variable real number that changes depending on the situation, or a character string that indicates a specific selection method. For example, when the top n people highly related to a KPI are to be used as role models, the role model selection conditions 207 are "top n people." When the KPIs of subjects belonging to a region are compared with the KPIs of subjects belonging to a region outside the region, and the region is set so that the difference between the two groups is maximized, the role model selection conditions 207 are "maximize the KPI difference between inside and outside the region." The role model selection conditions 207 may be set from the data input / output unit 101, or may be read from a setting file or the like that has been entered in advance. There is no limit to the number of conditions that can be set.

[0076] A specific example 403 of the role model selection conditions 207 includes a conditional expression for selecting subjects to be role models. The conditional expression used in the selection conditions is not limited to a particular method. In the example of FIG. 4C, the selection condition for role model condition pattern ID "0" is set to select the top 20% of subjects with a KPI of 1 or more as role models, and the selection condition for role model condition pattern ID "1" is set to select the top n subjects with a KPI of 5 or more as role models.

[0077] FIG. 4D is an explanatory diagram of the policy recipient information 208.

[0078] The policy recipient information 208 includes an ID that uniquely defines the setting of the policy recipient, identification information of the selection conditions used to set the policy recipient among the selection conditions set in the policy recipient selection conditions 209, identification information of the space used to set the policy recipient among the spaces set in the space information 205, and identification information of the subjects selected as policy recipients in the setting of the policy recipient among the subjects set in the subject information 204. Specifically, the policy recipient information 208 includes information of subjects who belong to a policy recipient area close to the role model area among the subjects projected onto a space whose axis is the feature indicated by the related feature 202.

[0079] A specific example 404 of the policy recipient information 208 includes a policy recipient ID, a policy recipient condition pattern ID, a space ID, and a target ID of the person who will be the target of the policy. The policy recipient condition pattern ID corresponds to the policy recipient condition pattern ID held in the policy recipient selection conditions 209. The space ID corresponds to the space ID held in the space information 205. The target ID corresponds to the target ID held in the target information 204. The policy recipient ID does not need to be expressed in the form of an integer, and may be expressed as a character string such as a hash value, as long as it corresponds to each policy recipient, the policy recipient selection conditions 209 when the person was selected as a policy recipient, and the space information 205.

[0080] FIG. 4E is an explanatory diagram of the policy recipient selection conditions 209.

[0081] The policy recipient selection conditions 209 include an ID (policy recipient condition pattern ID) that uniquely defines the setting of the selection conditions, and a selection condition that defines whether or not a target person is to be a policy recipient. The policy recipient selection conditions 209 may be a fixed real number, or a variable real number that changes depending on the situation. For example, if 30 people who are similar to role models are to be policy recipients, the policy recipient selection conditions 209 would be "30 people who are similar to role models." The policy recipient selection conditions 209 may be set from the data input / output unit 101, or may be read from a setting file or the like that has been entered in advance. There is no limit to the number of conditions that can be set.

[0082] A specific example 405 of the policy recipient selection conditions 209 includes a conditional expression for selecting subjects to be targeted by the policy. The conditional expression used in the selection conditions is not limited to a specific method. In the example of FIG. 4E, the selection condition for policy recipient condition pattern ID "0" is set to determine 30 subjects who are close to role models among subjects with a KPI of 1 or more as policy recipients, and the selection condition for policy recipient condition pattern ID "1" is set to determine subjects who are included in one square surrounding the role model area in space among subjects with a KPI of 2 or more as policy recipients.

[0083] FIG. 4F is an explanatory diagram of the integrated spatial information 210.

[0084] The integrated spatial information 210 is information that sets a space that integrates multiple spaces onto which the feature amounts of each subject are projected, and includes an ID that uniquely defines the spatial integration information, an ID of the condition set in the integrated spatial condition 211, and an ID of the space set in the spatial information 205. Specifically, the integrated spatial information 210 includes information on the space that is used as information to be provided when supporting policy planning.

[0085] A specific example 406 of the integrated spatial information 210 includes an integrated spatial information ID, a space integration condition pattern ID, a role model condition pattern ID, a policy target condition pattern ID, and a space ID. The space integration condition pattern ID corresponds to the space integration condition pattern ID held by the integrated spatial conditions 211. The role model condition pattern ID corresponds to the role model condition pattern ID held by the role model selection conditions 207. The policy target condition pattern ID corresponds to the policy applyee condition pattern ID held by the policy applyee selection condition ID 209. The space ID corresponds to the space ID held by the space information 205. The integrated spatial information ID does not need to be expressed in the form of an integer, as long as it corresponds to the integration conditions used when integrating each space with another space, and may be expressed as a character string such as a hash value.

[0086] FIG. 4G is an explanatory diagram of the integrated spatial condition 211.

[0087] The integrated space condition 211 includes an ID that uniquely defines the setting of the integration condition and a condition that defines whether or not to integrate the spaces set in the space information 205. The integrated space condition 211 may be a conditional expression that includes a fixed real number or a variable real number that changes depending on the situation, or may be a character string that indicates a specific selection method. For example, if the match rate of subjects belonging to the role model area in each space exceeds 80% and information on each space is provided as information for policy planning support, the integrated space condition 211 would be "80% match rate of subjects belonging to the role model area." The integrated space condition 211 may be set from the data input / output unit 101, or may be read from a setting file or the like that has been entered in advance. The number of conditions is not specified.

[0088] A specific example 407 of the integrated space condition 211 includes a conditional expression for determining whether to integrate spaces. The conditional expression used in the integration condition is not limited to a specific method. In the example of FIG. 4G, the integration condition of space integration condition pattern ID "0" indicates a condition that the two spaces are integrated if the matching rate of subjects belonging to the role model areas of the two spaces is 50% or more and the matching rate of subjects belonging to the policy recipient areas of the two spaces is 50% or more. On the other hand, the integration condition of space integration condition pattern ID "1" indicates a condition that the two spaces are integrated if the matching rate of subjects belonging to the role model areas of the two spaces is 80% or more.

[0089] FIG. 5A is an explanatory diagram of the policy information 212.

[0090] The policy information 212 includes an ID that uniquely defines the policy settings, the space set in the space information 205, the target person set in the target person information 204, and information required to support policy planning. A policy proposal associated with the axis of the space used to select the policy recipient is acquired from the feature table 200. By referring to the feasibility of the policy managed in the feature table 200, the space including the feature values ​​for which the policy can be considered and the policy recipient information are stored as the policy information 212. For example, in considering a policy for a new product campaign, if the axes of the space used to select the policy recipient are "number of visits" and "age," the policy proposal "campaign offering special benefits for visitors" associated with the feature value "number of visits" and "campaign for products targeted at the xx age group" associated with the feature value "age" are stored in the policy information 212. "xx" is determined by referring to the feature value of the role model.

[0091] A specific example 501 of the policy information 212 includes attributes such as a policy ID, a space ID, and a target ID. The space ID corresponds to the space ID held in the space information 205. If a policy is planned using information on multiple spaces, multiple space IDs are included in the attributes. The policy plan at that time may also be stored in the implemented policy. The policy ID does not need to be expressed in the form of an integer, and may be expressed as a character string such as a hash value, as long as it corresponds to each piece of information used when considering the policy. The attribute may be a discrete categorical variable, a continuous variable, or a character string.

[0092] FIG. 5B is an explanatory diagram of the transition information 213.

[0093] The transition information 213 includes information about the policy, information about the policy recipient held in the policy recipient information 208, information about indicators for confirming the effect of the policy, and information about the effect before and after the implementation of the policy.

[0094] A specific example 502 of the transition information 213 includes a policy ID, a policy applicator ID, an effect confirmation indicator pattern ID, and indicators before and after the implementation of the policy. The policy ID corresponds to the policy ID held in the policy information 212. The policy applicator ID corresponds to the policy applicator ID held in the policy applicator information 208. The effect confirmation indicator pattern ID corresponds to the effect confirmation indicator pattern ID held in the effect confirmation indicator information 214. The indicator may be a discrete categorical variable or a continuous variable as long as it can measure the effect of the policy, and the method is not limited. Furthermore, the indicator is not limited to one, and may be composed of several types of indicators.

[0095] FIG. 5C is an explanatory diagram of the effect confirmation index information 214.

[0096] The effect confirmation indicator information 214 includes an ID that uniquely defines the setting of an indicator when confirming the effect of a policy, and information about the evaluation indicator. The evaluation indicator represents a change in KPI before and after the implementation of a policy, position information of the policy recipient in a space based on the related feature amount 202 before and after the implementation of a policy, or a change in the similarity (or dissimilarity) between the role model and the policy recipient before and after the implementation of a policy. Here, the position information may be a position coordinate in space, or may be a discrete one-hot expression that indicates which category the person is in.

[0097] A specific example 503 of the effect confirmation indicator information 214 includes an effect confirmation indicator pattern ID and an indicator item used during evaluation. The effect confirmation indicator pattern ID does not need to be expressed in the form of an integer, and may be expressed as a character string such as a hash value, as long as it corresponds to each piece of information used during policy consideration. The attribute may be a discrete categorical variable, a continuous variable, or a character string.

[0098] FIG. 5D is an explanatory diagram of the setting information 215.

[0099] The setting information 215 includes information about the execution conditions and settings of the policy planning support system. Each item may be set via the data input / output unit 101, or the contents described in a setting file or the like may be read.

[0100] A specific example 504 of the setting information 215 includes a KPI feature ID, a KPI target, a policy ID, the number of spaces, the relevance, related feature selection conditions, role model selection conditions, policy adopter selection conditions, space integration conditions, and effect confirmation index items. The configuration of the setting information is not limited to the above. The KPI feature ID corresponds to the feature ID held in the feature table 200. The KPI target is stored in the KPI target of the KPI information 201.

[0101] The policy ID corresponds to the policy ID held in the policy information. If a policy ID and an effect confirmation indicator item are specified, the processing of the policy effect confirmation unit 111 is executed (corresponding to the case of No in step S1001 in FIG. 10 described later). If a policy ID is not specified, the feature selection unit 104 specifies a feature associated with the KPI feature ID (corresponding to the case of Yes in step S1001 in FIG. 10 described later). The decision as to whether to formulate a policy or check the effect of the policy may be made based on the presence or absence of a policy ID as described above, but it may also be made by providing separate information indicating whether it is time to formulate a policy and referring to that information.

[0102] Additionally, "-" may be used to represent a missing value where no value exists. Missing values ​​are not limited to "-", and can be any symbol that indicates the absence of a value for the corresponding feature, such as "Null" or "None", which are synonymous with "-". The number of spaces specifies the number of spaces to provide information when formulating policies. The role model selection conditions and related feature selection conditions are used by the role model selection unit 107 and the feature selection unit 104, respectively.

[0103] <Hardware configuration of the policy planning support system> FIG. 9 is a block diagram showing an example of the hardware configuration of the policy planning support system 100. As shown in FIG.

[0104] The policy planning support system 100 includes a processor 901 , a data storage device 902 , a communication device 903 , an input device 904 , an output device 905 , and a program storage device 906 .

[0105] The processor 901 is a processor such as a CPU (Central Processing Unit) and controls the policy planning support system 100 .

[0106] The data storage device 902 serves as a working area for the processor 901. The data storage device 902 is a non-transitory or temporary recording medium that stores various programs and data. Examples of the data storage device 902 include a read-only memory (ROM), a random access memory (RAM), a hard disk drive (HDD), a solid state drive (SSD), and a flash memory.

[0107] The communication device 903 connects to a network and transmits and receives data.

[0108] The input device 904 is a device for inputting data and information into the system, and examples of the input device 904 include a keyboard, a mouse, a touch panel, a numeric keypad, a scanner, and a microphone.

[0109] The output device 905 is a device that outputs data and information held by the system, and examples of the output device 905 include a display, a printer, and a speaker.

[0110] The program memory device 906 has a KPI selection unit 103, a feature selection unit 104, a mapping unit 105, a spatial integration unit 109, a policy selection unit 110, a policy effect confirmation unit 111, and a visualization unit 112, which are stored in the policy planning support system 100 of Figure 1.

[0111] The data input / output unit 101 and the program storage device 906, which include the KPI selection unit 103, feature selection unit 104, mapping unit 105, spatial integration unit 109, policy selection unit 110, policy effect confirmation unit 111, and visualization unit 112 shown in FIG. 1, are specifically realized by having the processor 901 execute a program stored in the data storage device 902. In other words, the program storage device 906 may be any of the storage devices constituting the data storage device 902. Furthermore, the database 102 shown in FIG. 1 is stored in the storage device of the policy planning support system 100 or an external computer.

[0112] <Overall system flowchart> FIG. 10 is a flowchart showing the overall processing procedure of the policy planning support system 100.

[0113] First, the policy planning support system 100 determines whether the system use is for policy planning (S1001). If it is for policy planning, step S1002 is executed next, and if not, step S1007 is executed next.

[0114] The KPI selection unit 103 acquires the setting information 215 and the feature amount table 200, and stores the feature amounts and feature amount targets in the setting information 215 in the KPI information 201 by referring to the KPI selection availability conditions in the feature amount table 200 (S1002). Details of this process will be described later with reference to FIG.

[0115] Next, the feature selection unit acquires the setting information 215, the KPI information 201, and the feature table 200, and acquires related feature amounts that are highly related to the selected KPI (S1003). Details of this process will be described later with reference to FIG.

[0116] Next, the mapping unit 105 generates a space with the related feature amount as an axis, acquires the feature amount of the subject from the subject information 204, and projects it onto the space (S1004). Details of this processing will be described later with reference to FIGS. 13 to 16.

[0117] Next, the spatial integration unit 109 acquires the setting information 215, the role model information 206 in a plurality of spaces, and the policy applicant information 208, and stores the information of the spaces that satisfy the conditions in the integrated spatial information 210 (S1005). Details of this process will be described later with reference to FIG. 17.

[0118] Next, the policy selection unit 110 acquires the policy information 212 corresponding to the relevant feature amount used for the axis of the space from the feature amount table 200 (S1006). Details of this process will be described later with reference to FIG. 18. Next, step S1008 is executed.

[0119] The policy effect confirmation unit 111 projects the feature amounts before and after the policy from the target person information 204 onto the space with the relevant feature amount used at the time of policy formulation as the axis, and calculates the transition information (S1007). Details of this process will be described later with reference to FIG. 19. Next, step S1008 is executed.

[0120] The visualization unit acquires the policy information 212 and the transition information 213, and provides the information necessary for policy formulation to the system user by displaying the spatial information 205 and the target person information 204 (S1008). Details of this process will be described later with reference to FIG. 20. <​​​​​​​​​​​​​​​​​​​ If it is possible to set it as a KPI, the KPI selection unit 103 stores the acquired feature amount and feature amount target in the KPI information 201 (S1104).

[0126] <Flowchart of feature selection section> FIG. 12 is a flowchart showing the processing procedure of the feature selection unit 104.

[0127] The feature quantity selection unit 104 acquires and sets the related feature quantity selection conditions 203 and the degree of association from the data input / output unit 101 or a setting file (S1201).

[0128] Next, the feature selection unit 104 acquires feature amounts from the feature amount table 200 (S1202).

[0129] Next, the feature amount selection unit 104 calculates the degree of association between the acquired feature amount and the KPI information 201 (S1203).

[0130] Next, the feature amount selection unit 104 determines whether the degree of association between the acquired feature amount and the KPI information 201 satisfies the related feature amount selection condition 203 (S1204).

[0131] If the degree of association between the acquired feature amount and the KPI information 201 satisfies the related feature amount selection condition 203, the feature amount selection unit 104 stores the acquired feature amount in the related feature amount 202 (S1205).

[0132] <Mapping flow chart> FIG. 13 is a flowchart showing the processing procedure of mapping section 105.

[0133] The mapping unit 105 sets the role model selection conditions 207 and the policy recipient selection conditions 209 based on information acquired from the data input / output unit 101 or the setting file (S1301).

[0134] Next, the mapping unit 105 generates a space whose axes are the features held in the related features 202 (S1302).

[0135] Next, the feature projection unit 106 of the mapping unit 105 acquires the feature of each subject from the subject information 204, and projects the feature of the subject onto a space whose axis is the related feature 202 (S1303). Details of this process will be described later with reference to FIG.

[0136] Next, the role model selection unit 107 of the mapping unit 105 selects role models based on the role model selection conditions 207 (S1304). Details of this process will be described later with reference to FIG.

[0137] Next, the policy recipient selection unit 108 of the mapping unit 105 selects a policy recipient based on the policy recipient selection conditions 209 (S1305). Details of this process will be described later with reference to FIG.

[0138] <Flowchart of feature projection part> FIG. 14 is a flowchart showing the processing procedure of the feature amount projection unit 106.

[0139] The feature projection unit 106 generates a space with the related feature 202 as an axis, the number of which is equal to the number of combinations of related feature (S1402).

[0140] Next, the feature projection unit 106 acquires the feature of each subject from the subject information 204 (S1403).

[0141] Next, the feature projection unit 106 projects the feature of the subject onto the generated space (S1404).

[0142] <Role Model Selection Department Flowchart> FIG. 15 is a flowchart showing the processing procedure of the role model selection unit 107.

[0143] The role model selection unit 107 sets the role model selection conditions 207 based on information acquired from the data input / output unit 101 or a setting file (S1501).

[0144] Next, the role model selection unit 107 acquires the projected feature amount of each subject (S1502).

[0145] Next, the role model selection unit 107 determines whether the acquired feature quantity satisfies the role model selection condition 207 (S1503).

[0146] If the acquired feature quantity satisfies the role model selection condition 207, the role model selection unit 107 stores the target person in the role model information 206 (S1504).

[0147] <Flowchart of the Policy Applicant Selection Department> FIG. 16 is a flowchart showing the processing procedure of the policy recipient selection unit 108.

[0148] The policy recipient selection unit 108 sets the policy recipient selection conditions 209 based on information acquired from the data input / output unit 101 or the setting file (S1601).

[0149] Next, the policy recipient selection unit 108 acquires the feature amount of the projected subject (S1602).

[0150] Next, the policy recipient selection unit 108 determines whether the acquired feature quantity satisfies the policy recipient selection condition 209 (S1602).

[0151] If the acquired feature quantity satisfies the policy recipient selection condition 209, the policy recipient selection unit 108 stores the target person in the role model information 206 (S1603).

[0152] <Spatial Integration Flowchart> FIG. 17 is a flowchart showing the processing procedure of the spatial integration unit 109.

[0153] The space integration unit 109 sets the integrated space condition 211 and a threshold value for the number of spaces to be generated based on information acquired from the data input / output unit 101 or a setting file (S1701).

[0154] Next, the space integration unit 109 checks whether the number of generated spaces exceeds the set threshold value for the number of spaces (S1702). If it does not exceed the threshold value, the process from S1702 to S1705 is repeated.

[0155] Next, the space integrating unit 109 acquires the space information 205 onto which the feature amount of the subject is projected, the role model information 206, and the policy recipient information 208 (S1703).

[0156] Next, the space integration unit 109 determines whether the acquired space satisfies the integrated space condition 211 (S1704).

[0157] If the acquired space satisfies the integrated space condition 211 , the space integration unit 109 stores the acquired space in the integrated space information 210 .

[0158] In step S1702, if the number of generated spaces exceeds the set threshold value for the number of spaces, the space integrating unit 109 ends the process.

[0159] <Flowchart of the Policy Selection Department> FIG. 18 is a flowchart showing the processing procedure of the feature amount projection unit 106.

[0160] The feature amount projection unit 106 acquires the spatial information 205 from the integrated spatial information 210 (S1801).

[0161] Next, the feature projection unit 106 acquires related feature values ​​that are set as axes in the acquired space (S1802).

[0162] Next, the feature projection unit 106 determines whether the acquired related feature 202 is included in the target of the measure (S1803).

[0163] If the acquired related feature 202 is not included in the target of the measure, the process ends here.

[0164] If the acquired related feature 202 is included in the target of the policy, the feature projection unit 106 stores the spatial information and the policy recipient ID in the policy information 212 and provides the acquired spatial information to the system user via the visualization unit (S1804).

[0165] Next, the feature projection unit 106 determines whether or not there is a proposed measure associated with the feature in advance in the feature table 200 (S1805).

[0166] If there is no measure plan previously associated with a feature in the feature table 200, the process ends here.

[0167] Next, the feature projection unit 106 checks whether the system user has come up with a new measure based on the provided measure (S1806).

[0168] When a system user comes up with a new policy plan based on the provided policy plan, the feature projection unit 106 inputs the new implementation plan via the data input / output unit 101, and the feature projection unit 106 stores the input implementation plan in the implementation policy of the policy information 212 (S1807).

[0169] If the system user does not devise a new policy plan based on the provided policy plan, but instead adopts the implementation plan provided by the system as is, the feature projection unit 106 stores the policy plan that has been previously associated with the feature in the implementation policy of the policy information 212 (S1808).

[0170] <Flowchart of the Policy Effectiveness Verification Department> FIG. 19 is a flowchart showing the processing procedure of the policy effect confirmation unit 111.

[0171] The policy effect confirmation unit 111 sets the policy information 212 and the effect confirmation indicator information 214 based on information acquired from the data input / output unit 101 or a setting file (S1901).

[0172] Next, the policy effect confirmation unit 111 acquires the space information stored in the policy information 212, and creates a new space with the same axes (S1902).

[0173] Next, the policy effect confirmation unit 111 acquires the feature amount of the subject from the subject information 204 stored in the policy information 212 (S1903).

[0174] Next, the policy effect confirming unit 111 projects the acquired pre-policy related feature amount of the subject onto the generated space (S1904).

[0175] Next, the policy effect confirming unit 111 projects the acquired feature amount of the subject after the policy is implemented onto the newly generated space (S1905).

[0176] Next, the policy effect confirmation unit 111 calculates an effect confirmation index based on the information before and after the policy for each subject, and stores the calculated index in the transition information 213 (S1906).

[0177] <Flowchart of visualization part> FIG. 20 is a flowchart showing the processing procedure of the visualization unit 112.

[0178] The visualization unit 112 is capable of visualizing all parameters held by the policy planning support system 100, and as an example, a flowchart for visualizing parameters generated by the processing of the policy selection unit 110 is described here.

[0179] The visualization unit 112 sets the policy information 212 based on the information acquired from the data input / output unit 101 or the setting file (S2001).

[0180] Next, the visualization unit 112 acquires the integrated spatial information 210 and the target person information 204 from the policy information 212 (S2002).

[0181] Next, the visualization unit 112 acquires the spatial information 205 from the integrated spatial information 210, and projects the acquired related feature amount of the subject onto the acquired space (S2003).

[0182] Next, the visualization unit 112 provides the information on the spatial axes and the target person information to the system user (S2004). [Example]

[0183] <Example 1: Supermarket Campaign> 21A to 24C, a first example will be described in which the policy planning support system 100 is used in a supermarket campaign. The configuration of the system according to the first example is the same as that of the system according to the embodiment shown in FIGS. 1 to 20, except for the differences described below, and therefore a description thereof will be omitted.

[0184] In Example 1 for a supermarket campaign, a campaign is implemented to improve KPIs while taking into consideration the preferences and characteristics of customers. Since many campaigns are implemented, such as for new products and seasonal limited-edition products, it is necessary to appropriately implement campaigns by narrowing down the target audience to customers who are expected to be effective, so as not to lose the sense of specialness of the campaign by frequently implementing campaigns to all customers. Furthermore, even if there has been no history of running a campaign, it is necessary to implement an effective campaign by identifying customers who are expected to be effective.

[0185] FIG. 21A is an explanatory diagram of the feature amount table 200 in the first embodiment.

[0186] A specific example 2101 of the feature amount table 200 in FIG. 21A is a different example from the specific example 300 of the feature amount table 200 shown in FIG. 3A, and shows an example of the feature amount table 200 in an example of a supermarket campaign.

[0187] A specific example 2101 of the feature table 200 holds, as feature names, "number of visits," "average number of items purchased per visit," "number of product A purchases," "number of product B purchases," "average time of purchase," "gender," "age," "place of residence," etc. Of these, "gender," "age," and "place of residence" are demographic data (i.e., demographic: True) and are set so that they cannot be selected as KPIs (i.e., KPI setting availability: False). On the other hand, "number of visits," "average number of items purchased per visit," "number of product A purchases," "number of product B purchases," and "average time of purchase" are not demographic data and are set so that they can be selected as KPIs.

[0188] Furthermore, for feature quantities other than "residence", it is set that measures can be formulated (i.e., measure possibility: True), and one or more corresponding measure proposals are stored. In the example of FIG. 21A, "visit bonus" and "stamp card" are stored as measure proposals corresponding to "number of visits". These are set in advance because it is considered effective to provide a visit bonus and a stamp card to increase the number of visits by the target person (in this example, supermarket customers). These can be set arbitrarily by the system user as they deem effective, and may be set, for example, based on experience or past performance.

[0189] Similarly, in the example of FIG. 21A, "bulk buying" (e.g., discounts for bulk purchases) is set as a proposed measure corresponding to the "average number of items purchased per purchase." Price reductions for each item and the provision of special benefits to purchasers of each item are set as proposed measures corresponding to the "number of purchases of item A" and the "number of purchases of item B." Time sales are set as a proposed measure corresponding to the "average purchase time period." Consideration of preferences by gender (e.g., offering popular items according to gender) is set as a proposed measure corresponding to "gender." Consideration of preferences by age group (e.g., offering popular items according to age) is set as a proposed measure corresponding to "age."

[0190] Note that the KPI selection availability and the implementation availability may be set in accordance with the intention of campaign management. For example, in the specific example 2101 of the feature table 200, the implementation availability for gender is set to True, but if there is no intention to implement a campaign focusing on gender, the option can be intentionally set to False, and the gender item will be excluded from the information for implementing measures provided by the measure planning support system 100.

[0191] FIG. 21B is an explanatory diagram of the setting information 215 in the first embodiment.

[0192] 21B is a different example from the specific example 504 of the setting information 215 shown in FIG. 5D, and shows an example of the setting information 215 at the time of policy planning in an embodiment of a supermarket campaign. In this example, the specific example 2102 of the setting information 215 is expressed as a setting file, and the setting file is read when the program is executed. However, the setting information 215 does not need to be specified by the setting file, and may be specified via the data input / output unit 101.

[0193] In the example of FIG. 21B, "2" is stored as the KPI feature ID. This indicates that the feature "number of purchases of product A" in the example of FIG. 21A is set as the KPI. Then, "increasing" the number of purchases of product A is set as the KPI target, "2" is set as the number of spaces, and "importance" is set as the relevance. Then, a related feature selection condition is set such that features whose relevance (i.e., importance) with the KPI is 0.2 or greater and whose value is greater than 0 are selected as related features. In addition, a condition is set such that the top three subjects with the highest number of purchases of product A are selected as role models, and four subjects around the role model area are selected as policy recipients. Furthermore, a condition is set such that two spaces are merged in descending order of correlation with the KPI if the matching rate of subjects belonging to each role model area is 30% or greater.

[0194] FIG. 21C is an explanatory diagram of the KPI information 201 in the first embodiment.

[0195] A specific example 2103 of KPI information 201 in Fig. 21C is a different specific example from the specific example 301 of KPI information 201 shown in Fig. 3B, and shows an example of KPI information 201 in an example embodiment of a supermarket campaign. The KPI information 201 is referenced from the setting information 215, and is assigned a KPI ID before being stored.

[0196] In the example of FIG. 21C, as shown in the specific example 2102 of the setting information 215 in FIG. 21B, increasing the number of purchases of product A is stored in association with the KPI ID "0."

[0197] FIG. 22A is an explanatory diagram of the subject information 204 in the first embodiment.

[0198] 22A is a different example from the specific example 304 of the target person information 204 shown in Fig. 3E, and shows an example of target person information 204 in an embodiment of a supermarket campaign. Specifically, the specific example 2201 of the target person information 204 shown in Fig. 22A holds the actually measured values ​​of each feature (i.e., number of visits, average number of items purchased per visit, number of purchases of product A, number of purchases of product B, average time period for purchase, etc.) of the specific example 2101 of the feature table 200 shown in Fig. 21A for each target person.

[0199] FIG. 22B is an explanatory diagram of the related feature amount 202 in the first embodiment.

[0200] 22B is a different specific example from the specific example 302 of the related feature 202 shown in FIG. 3C, and shows an example of the related feature 202 in an embodiment of a supermarket campaign. The specific example 2202 of the related feature 202 shown in FIG. 22B stores only the feature quantities that satisfy the related feature selection conditions of the specific example 2102 of the setting information 215 shown in FIG. 21B (i.e., the degree of relevance with the KPI is 0.2 or more and the feature quantity value is greater than 0). That is, in this example, the number of visits, the average number of items purchased per visit, the number of purchases of product B, the average time period for purchase, and the age are stored as feature quantities whose relationship with the KPI, i.e., the number of purchases of product A, satisfies the above conditions.

[0201] FIG. 22C is an explanatory diagram of the related feature selection condition 203 in the first embodiment.

[0202] A specific example 2203 of the related feature selection conditions 203 in Figure 22C is a different example from the specific example 303 of the related feature selection conditions 203 shown in Figure 3D, and shows an example of the related feature selection conditions 203 in an embodiment of a supermarket campaign. The related feature selection conditions 203 are specified by the setting information 215. In the example of Figure 22C, a condition is specified that the degree of relevance with the KPI is 0.2 or more and the feature value is greater than 0, and is stored as the related feature 202. The degree of relevance is calculated based on the correlation specified in the setting information 215 (importance in the example of Figure 21B).

[0203] FIG. 22D is an explanatory diagram of the spatial information 205 in the first embodiment.

[0204] A specific example 2204 of the space information 205 in FIG. 22D is a different specific example from the specific example 401 of the space information 205 shown in FIG. 4A, and shows an example of the space information 205 in an example of a supermarket campaign. The space information 205 stores KPI information 201 used when generating the space, related feature selection conditions 203, and related feature IDs that serve as the axes of the space. Here, the related feature IDs correspond to the related feature IDs of the specific example 2202 of the related feature 202 shown in FIG. 22B. That is, in the example of FIG. 22D, two spaces are set, and the number of visits to the store is set on one axis of the space with a space ID of "0" and the average number of purchases per visit is set on the other axis, while the number of purchases of product B is set on one axis of the space with a space ID of "1" and the average time period of purchase is set on the other axis.

[0205] FIG. 23A is an explanatory diagram of the role model information 206 in the first embodiment.

[0206] A specific example 2301 of the role model information 206 in FIG. 23A is a different example from the specific example 402 of the role model information 206 shown in FIG. 4B, and shows an example of the role model information 206 in an example of a supermarket campaign.

[0207] FIG. 23B is an explanatory diagram of the role model selection conditions 207 in the first embodiment.

[0208] 23B is a different specific example from the specific example 403 of the role model selection conditions 207 shown in Fig. 4C, and shows an example of the role model selection conditions 207 in an example embodiment of a supermarket campaign. The specific example 2302 of the role model selection conditions 207 is specified in the setting information 215, and shows a condition that the area including the top three people with the highest KPIs among the mapped subjects is set as the role model area.

[0209] FIG. 23C is an explanatory diagram of the policy recipient information 208 in the first embodiment.

[0210] The specific example 2303 of policy recipient information 208 in Figure 23C is a different specific example from the specific example 404 of policy recipient information 208 shown in Figure 4D, and shows an example of policy recipient information 208 in an example of a supermarket campaign.

[0211] FIG. 23D is an explanatory diagram of the policy recipient selection conditions 209 in the first embodiment.

[0212] 23D is a different example from the specific example 405 of the policy recipient selection conditions 209 shown in Fig. 4E, and shows an example of the policy recipient selection conditions 209 in an example embodiment of a supermarket campaign. The specific example 2304 of the policy recipient selection conditions 209 shows a condition that, among the subjects specified in the setting information 215 and mapped, an area including four people who are close to the role model area is set as the policy recipient area.

[0213] FIG. 23E is an explanatory diagram of the integrated spatial information 210 in the first embodiment.

[0214] The specific example 2305 of integrated spatial information 210 in Figure 23E is a different example from the specific example 406 of integrated spatial information 210 shown in Figure 4F, and shows an example of integrated spatial information 210 in an example of a supermarket campaign.

[0215] FIG. 23F is an explanatory diagram of the integrated spatial condition 211 in the first embodiment.

[0216] 23F is a different specific example from the specific example 407 of the integrated space condition 211 shown in FIG. 4G, and shows an example of the integrated space condition 211 in an example of a supermarket campaign. The specific example 2306 of the integrated space condition 211 shows a condition that, when the matching rate of subjects belonging to the role model area among multiple spaces specified in the setting information 215 and based on related feature quantities is 30% or more, information about the space is also provided when policy planning is carried out.

[0217] In this example, information that integrates two spaces that are highly correlated with the KPI is provided to the system user in accordance with the number of spaces and the integrated space conditions specified in the specific example 2102 of the setting information 215. Since the policy information 212 is output based on the related feature amounts that serve as the axes of multiple spaces that are highly correlated with the KPI, the policy planning support system 100 can provide information on multiple related feature amounts and information on policies that are expected to have an effect on improving the KPI.

[0218] FIG. 24A is an explanatory diagram of the visualization unit 112 according to the first embodiment.

[0219] Display example 2401 in Figure 24A shows an example of a display by the visualization unit 112 in an embodiment of a supermarket campaign. Space 2402 and space 2403 indicate the two spaces set in embodiment 1. Star 2404 indicates subjects selected as policy recipients in both spaces, and circle 2405 indicates other subjects. Dashed line 2406 indicates the dividing line between the role model areas, and dashed-dotted line 2407 indicates the dividing line between the policy recipient areas. Pop-out 2408 shows an example of what is displayed when a system user clicks on the GUI, allowing detailed information and the like to be confirmed.

[0220] FIG. 24B is an explanatory diagram of the policy information 212 in the first embodiment.

[0221] 24 is a different example from the specific example 501 of the policy information 212 shown in FIG. 5A, and shows an example of the policy information 212 in an embodiment of a supermarket campaign. The policy information 212 stores detailed settings for implementing the policy and the target IDs of the persons to whom the policy applies. The actually implemented policy may be recorded via the data input / output unit 101 as the implemented policy.

[0222] FIG. 24C is an explanatory diagram of a display example of the transition information 213 in the first embodiment.

[0223] Display example 2410 of transition information 213 in Figure 24C shows an example of displaying transition information 213 in an example of a supermarket campaign, and includes a specific example 2411 of setting information 215, transition examples 2412 and 2413 of features in each space, and a specific example 2415 of transition information 213.

[0224] A specific example 2411 of the setting information 215 shows an example of the setting information 215 for confirming the effectiveness of a policy. When confirming the effectiveness of a policy, information is acquired and stored from the policy ID in order to target the axis and target persons at the time of implementing the policy.

[0225] Transition example 2412 shows an example of transition of the feature amounts of a policy recipient before and after the implementation of a policy in a space where one axis is the number of visits to the store and the other axis is the average number of items purchased per visit. Transition example 2413 shows an example of transition of the feature amounts of a policy recipient before and after the implementation of a policy in a space where one axis is the number of purchases of product B and the other axis is the average time period for one purchase. Arrow 2414 indicates the transition before and after the policy. In other words, the feature amounts of a certain subject before and after the implementation of the policy are displayed at the start and end of arrow 2414, respectively.

[0226] Specific example 2415 of transition information 213 is a different specific example from specific example 502 of transition information 213 shown in FIG. 5B, and shows in a table the transition of feature values ​​for each policy recipient before and after the policy is implemented. When the related feature values ​​202 for each subject are projected onto a continuous space, the effectiveness of the campaign can be quantitatively evaluated based on how close the distance between the role model or the center of the role model area and the feature values ​​of each subject has become due to the implementation of the policy. For example, the number of subjects belonging to the role model area before and after the policy is implemented may be compared, and the effectiveness of the campaign may be evaluated based on the increase, or the effectiveness of the campaign may be evaluated based on the increase in the number of subjects who satisfy role model selection conditions 207.

[0227] In the above-described first embodiment, when the KPI is set to "number of purchases of product A," the related features identified include "number of visits," "average number of items purchased per visit," "number of purchases of product B," "average time period for purchase," and "age." In this case, for example, a five-dimensional space having axes corresponding to the respective related features can be generated, and the features of each subject can be projected onto that space. However, in this embodiment, multiple spaces with three or fewer dimensions (preferably two or fewer dimensions as shown in FIG. 24A ) are generated, and the features of each subject are projected onto each space, making it possible to display the space onto which the features are projected in a form that can be visually grasped by the system user.

[0228] When multiple spaces are generated in this way and features are projected in each, policy recipients may be selected in each space and policies corresponding to the related features set as axes of each space may be implemented for the policy recipients. On the other hand, the spaces may be integrated when a predetermined space integration condition is met, such as the matching rate of subjects included in a predetermined area in each space (for example, the role model area or the policy recipient area) is higher than a predetermined standard.

[0229] For example, subjects who are selected as policy recipients in all of the multiple integrated spaces are more similar to role models in terms of related features than subjects who are not, and it is expected that applying policies will be more effective. Therefore, when spaces are integrated, subjects who are included in the policy recipient area in all of the multiple spaces may be ultimately selected as policy recipients, and policies corresponding to the axes of the integrated spaces may be implemented.

[0230] Furthermore, when integrating spaces, the degree of relevance between the related feature set as the axis of each space and the KPI may be referenced. For example, among subjects who are in the policy recipient area in one space and not in the policy recipient area in the other space, subjects who are in the policy recipient area in the space where the related feature set as the axis has a higher relevance to the KPI may be selected as policy recipients.

[0231] Alternatively, spaces with a low matching rate of subjects may be merged. In this case, different policy recipients are selected in each space, and by implementing policies corresponding to the axes of each space for each policy recipient, it becomes possible to implement multiple policies in parallel for different subjects without mutual interference.

[0232] The same applies to the following Examples 2 and 3. An example of integrating spaces with a low matching rate of subjects will be described in Example 3. [Example]

[0233] <Example 2: Activating learning content> 25A to 27B, a second example will be described in which the policy planning support system 100 is used to activate learning content. The configuration of the system according to the second example is the same as that of the system according to the embodiment shown in FIGS. 1 to 20, except for the differences described below, and therefore a description thereof will be omitted.

[0234] In Example 2 of revitalizing learning content, measures are implemented to improve KPIs while taking into consideration the items (subjects, skills, etc.) that the target person wants to learn and the learning materials (videos, articles, books, etc.) that they want to learn. Excessive measures may reduce motivation to learn, so in order to encourage as much voluntary learning as possible, it is necessary to implement effective measures by focusing on target people who are expected to benefit. Even if there is no history of implementing measures, it is necessary to implement effective measures while identifying target people who are expected to benefit.

[0235] FIG. 25A is an explanatory diagram of the feature amount table 200 in the second embodiment.

[0236] Specific example 2501 of feature table 200 in FIG. 25A is a different specific example from specific example 300 of feature table 200 shown in FIG. 3A, and shows an example of feature table 200 in an embodiment of activating learning content.

[0237] A specific example 2501 of the feature table 200 holds feature names such as "number of content completions," "number of article content views," "number of video content views," "age," "occupation," and "job title." Of these, "age," "occupation," and "job title" are demographic data and are set to be unselectable as KPIs. On the other hand, "number of content completions," "number of article content views," and "number of video content views" are not demographic data and are set to be selectable as KPIs.

[0238] Furthermore, for feature quantities other than "age," it is set that measures can be planned, and one or more corresponding measure proposals are stored. In the example of FIG. 25A, measures corresponding to the "number of content completions" are stored, such as encouraging the user to complete the content they are currently viewing and introducing content that may be of interest. These measures are set in advance because they are considered to be effective in increasing the number of content completions by the target user.

[0239] Similarly, in the example of FIG. 25A, the proposed measure corresponding to the "number of views of article content" is set to introduce article-type content. The proposed measure corresponding to the "number of views of video content" is set to introduce video-type content. The proposed measure corresponding to "occupation" is set to introduce content related to some skill (for example, skills required for the target occupation). The proposed measure corresponding to "job position" is set to introduce content suitable for the job position.

[0240] Here, the number of views of article content and the number of views of video content are given as examples, but the types of content are not limited to these. The number and types of features are also not limited to these, and there are no restrictions on the content as long as they can be acquired as data. As with Example 1, the availability of KPIs and measures may be set according to the wishes of the learning content administrator.

[0241] FIG. 25B is an explanatory diagram of the setting information 215 in the second embodiment.

[0242] Specific example 2502 of setting information 215 in Figure 25B is a different example from specific example 504 of setting information 215 shown in Figure 5D, and shows an example of setting information 215 in an embodiment of activating learning content. In this example, specific example 2502 of setting information 215 is represented as a setting file, and this setting file is read when the program is executed. However, setting information 215 does not need to be specified by a setting file, and may be specified via data input / output unit 101.

[0243] In the example of Figure 25B, "0" is stored as the KPI feature ID. This indicates that the feature "number of content completions" in the example of Figure 25A is set as the KPI. Furthermore, "increasing" the number of content completions is set as the KPI goal, the number of spaces is set to "20," and the relevance is set to "SHAP (SHapley Additive exPlanations) value." This indicates the importance (magnitude of contribution) of each feature during learning or inference of a machine learning model, etc. In this example, it means the importance of each feature to the KPI. Furthermore, a related feature selection condition is set such that features with a relevance to the KPI of 0.2 or greater and a value greater than 0 are selected as related features. Furthermore, a condition is set such that the space is divided into cells of a specified size, and subjects belonging to the cell with the highest average KPI within the cells (role model area) are selected as role models, and subjects belonging to cells surrounding the role model area (e.g., cells adjacent to the role model area) are selected as policy recipients. Furthermore, a condition is set that the two spaces will be merged if the matching rate of subjects belonging to each role model area is 100%.

[0244] FIG. 26 is an explanatory diagram of selection of policy recipients in the second embodiment.

[0245] Display example 2601 in Figure 26 shows an example in which role model areas and policy recipient areas are defined for multiple spaces centered on related features 202 stored by the feature selection unit 104 based on role model selection conditions 207 and policy recipient selection conditions 209, and subjects belonging to the role model areas and policy recipient areas are picked out in accordance with integrated space conditions 211.

[0246] Space 2602 shows multiple spaces centered around the related feature 202, and spaces 2603 and 2606 show spaces within space 2602 that satisfy the integrated space condition 211. The area surrounded by dotted lines represents the role model area, and the area surrounded by dashed lines represents the policy recipient area. Stars 2604 represent subjects who appear in common in each space, and circles 2605 represent other subjects.

[0247] For subjects in the policy target area that appear in common in each space, information necessary for policy planning, such as information on the related feature 202 used on the spatial axis and policy proposals based on that information, is output. By using spatial axes with a high matching rate for subjects, it is possible to obtain a wide range of information from the related feature 202 that takes into account the common characteristics of each subject.

[0248] In this example, two spaces are picked up, but since the number of spaces is set to 20 in the specific example 2502 of the setting information 215, a maximum of 20 spaces may be picked up.

[0249] In Fig. 26, the space is divided into squares of a predetermined size, as in space 2603, and is expressed in a discrete space that shows how many people fit each square. Space 2607 expresses only the number of people fit in space 2603. By expressing in a discrete space, it is possible to measure the effectiveness of a policy by comparing the number of people fit, without considering the scale of the values ​​of each feature.

[0250] FIG. 27A is an explanatory diagram of a display example of the transition information 213 in the second embodiment.

[0251] Display example 2701 of transition information 213 in Figure 27A shows an example of confirming the effects before and after implementing measures in an embodiment of activating learning content, and includes specific example 2702 of setting information 215, transition examples 2703 and 2704 of features in each space, and specific example 2705 of transition information 213.

[0252] Specific example 2702 of setting information 215 shows setting information 215 at the time of confirming the effectiveness of a policy. Examples 2703 and 2704 of transitions of feature amounts in each space show examples of transitions of feature amounts of a policy recipient before and after implementation of the policy. As in Example 1, the arrows in the figure indicate transitions before and after the policy. In this example, the effectiveness of the policy is evaluated based on the effect confirmation index items in specific example 2702 of setting information 215 by checking whether a subject who did not belong to the role model area before implementation of the policy has transitioned to belong to the role model after implementation of the policy.

[0253] A specific example 2705 of the transition information 213 shows in a table whether a subject who was not in the role model domain before the implementation of the policy transitioned to the role model domain after the implementation of the policy as the transition information 213. By performing an evaluation in a discrete space, the effect of the policy can be evaluated regardless of the scale of the value of each feature amount.

[0254] FIG. 27B is an explanatory diagram of editing the feature amount table 200 in the second embodiment.

[0255] Editing screens 2706 and 2707 in FIG. 27B are diagrams showing an operation for editing information in the feature table 200 via the data input / output unit 101 and the visualization unit 112. Editing screen 2706 shows the screen before the operation, and editing screen 2707 shows the screen after the editing. In this example, the measure feasibility of the feature "number of views of article content" is edited from True to False. As a result, the next time the measure planning support system 100 is used to provide measure planning support, the feature "number of views of article content" will not be selected as the related feature 202. When the measure planning support system 100 is used continuously, it becomes possible to support measure planning in line with the system user's wishes. [Example]

[0256] <Example 3: Displaying Web Advertisements> 28A to 29B, a third embodiment will be described in which the policy planning support system 100 is used to display a web advertisement. The configuration of the system according to the third embodiment is the same as that of the system according to the embodiment shown in FIGS. 1 to 20, except for the differences described below, and therefore the description will be omitted.

[0257] In Example 3 for displaying web advertisements, similar to Example 1, it is necessary to display advertisements that improve KPIs while taking into consideration the preferences and characteristics of users. Displaying many web advertisements or displaying web advertisements too frequently can cause discomfort to the target user and may reduce the number of visits to the website that displays the advertisements. Therefore, it is necessary to display web advertisements at a timing and frequency appropriate for the target user, who is expected to see an improvement in KPIs.

[0258] FIG. 28A is an explanatory diagram of a Web advertisement in the third embodiment.

[0259] Advertisement example 2801 in Figure 28A is an example of a pattern of web advertising. Screen 2802 is an example of a website screen displaying a banner advertisement. In this example, the advertisement does not affect the viewing of the website itself, but is not easily noticed. Screen 2803 is an example of a website screen displaying a pop-up advertisement. In this example, the advertisement is sure to catch the eye, but it interferes with the viewing of the website, so it is not desirable to display it frequently. There are other patterns of web advertising, but the type is not limited as long as it is a web advertisement. Example 3 for displaying web advertisements supports the planning of measures that include not only the content of the web advertisements but also the patterns of the web advertisements.

[0260] FIG. 28B is an explanatory diagram of the feature amount table 200 in the third embodiment.

[0261] The specific example 2804 of the feature amount table 200 in FIG. 28B is a different example from the specific example 300 of the feature amount table 200 shown in FIG. 3A, and shows an example of the feature amount table 200 in an embodiment of displaying a Web advertisement.

[0262] A specific example 2804 of the feature table 200 holds feature names such as "Number of views of XX website," "Viewing time of XX website," "Total number of ad clicks," "Number of banner ad clicks," "Number of pop-up ad clicks," "Age," and "Address." Of these, "Age" and "Address" are demographic data and are set not to be selectable as KPIs. On the other hand, "Number of views of XX website," "Viewing time of XX website," "Total number of ad clicks," "Number of banner ad clicks," and "Number of pop-up ad clicks" are not demographic data and are set to be selectable as KPIs.

[0263] Furthermore, for feature quantities other than "total number of ad clicks," measures are set to be possible, and one or more corresponding measure proposals are stored. In the example of Figure 28B, a measure proposal corresponding to "number of views of X website" is stored, which is to display products related to the website. This is set in advance because it is thought to be effective in increasing the number of times the target person views the website.

[0264] Similarly, in the example of FIG. 28B, a proposed measure corresponding to "time spent browsing a certain website" is set to reduce the frequency of advertisements for people who browse for a long time and to increase the frequency of advertisements for people who browse for a short time. A proposed measure corresponding to "number of banner ad clicks" is set to preferentially display banner ads to people who click on banner ads frequently and to preferentially display a different pattern of advertisements to people who click on banner ads few times. A proposed measure corresponding to "number of pop-up ad clicks" is set to preferentially display pop-up ads to people who click on pop-up ads frequently and to preferentially display a different pattern of advertisements to people who click on pop-up ads few times. A proposed measure corresponding to "occupation" is set to introduce content related to some skill (for example, a skill required for a target occupation). A proposed measure corresponding to "age" is set to prepare advertisements for each generation. A proposed measure corresponding to "address" is set to display advertisements for facilities near the person's place of residence.

[0265] The feature quantities "number of website visits" and "viewing time" are obtained from cookies. "Age" and "Address" are obtained from account information for subjects who have registered an account. For subjects who do not have an account, these are treated as missing values ​​such as "Null" or "-".

[0266] FIG. 29A is an explanatory diagram of the setting information 215 in the third embodiment.

[0267] Specific example 2901 of setting information 215 in FIG. 29A is a different specific example from specific example 504 of setting information 215 shown in FIG. 5D, and shows an example of setting information 215 in an embodiment of displaying a Web advertisement.

[0268] In the example of FIG. 29A, "0" is stored as the KPI feature ID. This indicates that the feature "Number of views of X website" in the example of FIG. 28B is set as the KPI. Furthermore, "increasing" the number of views of the website is set as the KPI target, the number of spaces is set to "2," and the relevance is set to "correlation." Furthermore, a related feature selection condition is set such that features whose relevance to the KPI is 0.2 or greater and whose value is greater than 0 are selected as related features. Furthermore, a condition is set such that the space is divided into cells of a predetermined size, and subjects belonging to the cell with the highest average KPI within the cell (role model area) are selected as role models, and subjects belonging to cells surrounding the role model area (e.g., cells adjacent to the role model area) are selected as policy recipients. Furthermore, a condition is set such that two spaces are merged if the match rate of subjects belonging to each policy recipient area is 0% and the match rate of subjects belonging to each role model area is 0%. In this example, the spatial integration conditions are set to a 0% match rate for subjects belonging to the role model area and a 0% match rate for subjects belonging to the policy recipient area, allowing for different measures to be considered for completely different subjects.

[0269] FIG. 29B is an explanatory diagram of selection of policy recipients in the second embodiment.

[0270] Display example 2902 in Figure 29B shows an example in which role model areas and policy recipient areas are defined for multiple spaces centered on related features 202 stored by the feature selection unit 104 based on role model selection conditions 207 and policy recipient selection conditions 209, and spaces with a low matching rate of subjects belonging to the role model area and policy recipient area are picked out.

[0271] Among the multiple spaces, a subject represented by a rectangle 2903 in space 2905 is selected as a role model in space 2905, but is not selected as either a role model or a policy recipient in another space, space 2906. On the other hand, a subject represented by a triangle 2904 in space 2906 is selected as a role model in space 2906, but is not selected as either a role model or a policy recipient in space 2905. By picking out spaces with a low matching rate of subjects in this way, it is possible to reduce interference between policies, and therefore it is possible to implement different policies in parallel based on the axes of multiple spaces for each policy recipient in the policy recipient areas in multiple spaces.

[0272] Other specific methods of the policy planning support system 100 in the third embodiment are similar to those in the first and second embodiments.

[0273] Furthermore, the system according to the embodiment of the present invention may be configured as follows.

[0274] (1) A policy planning support system, comprising a processor (e.g., processor 901) and a storage device (e.g., data storage device 902 and program storage device 906), wherein the storage device holds subject information (e.g., subject information 204) including a plurality of feature amounts related to each of a plurality of subjects, and the plurality of feature amounts include at least one of feature amounts related to the behavior of the subjects and feature amounts related to the attributes of the subjects, and the processor identifies, among the plurality of feature amounts, two or more feature amounts related to a feature amount designated as an evaluation index (e.g., KPI) as related feature amounts (e.g., For example, step S1003), the related feature of each of the subjects is projected onto a space having each of the two or more related feature as an axis (for example, step S1004), and one or more of the subjects whose evaluation level based on the feature specified as the evaluation index meets a predetermined standard is selected as a role model (for example, step S1005, FIG. 15). Information for displaying the space onto which the related feature of the multiple subjects including the role model is projected is output as information to support the formulation of measures to improve the evaluation index (for example, step S1008, FIG. 20).

[0275] This makes it possible to provide information for planning measures that are expected to be effective, even in cases where there has been no history of implementing such measures. Specifically, by displaying feature quantities that are highly correlated with KPIs for multiple subjects, including role models, in a visually comparable format, it is possible to support the planning of measures to be implemented and the selection of subjects to whom those measures should be applied.

[0276] (2) In the policy planning support system described in (1) above, the processor identifies an area in the space that includes the position where the feature of the subject selected as the role model is projected as the role model area (e.g., step S1005, Figure 15), identifies an area within a predetermined range from the role model area as the policy recipient area (e.g., step S1005, Figure 16), selects one or more subjects corresponding to one or more of the feature projected into the policy recipient area as policy recipients (e.g., step S1005, Figure 16), and outputs information about the policy recipients as information to support the policy planning (e.g., step S1008, Figure 20).

[0277] This makes it possible to narrow down the target population for whom the measures are likely to be effective.

[0278] (3) In the policy planning support system described in (2) above, the processor generates two or more combinations of the related features, and for each of the two or more combinations of the related features, generates a space onto which the related features of each subject are projected (e.g., step S1004, Figure 14), and selects the policy recipient in each of the spaces.

[0279] This makes it possible to display the subject's features in a visually comparable manner even when a large number of related features are identified.

[0280] (4) In the policy planning support system described in (3) above, the processor selects the policy recipient based on the degree of relevance between the related features set as axes in each of the spaces and the evaluation index.

[0281] This makes it possible to appropriately select subjects for whom measures are likely to be effective even when a large number of related features are identified.

[0282] (5) In the policy planning support system described in (4) above, the processor outputs information about subjects selected as policy recipients in common across multiple spaces as information to support the policy planning.

[0283] This makes it possible to appropriately select subjects for whom measures are likely to be effective even when a large number of related features are identified.

[0284] (6) In the policy planning support system described in (5) above, the processor outputs, as information to support the policy planning, information regarding a subject who has been selected as the policy recipient in at least one of the multiple spaces and has not been selected as the policy recipient in at least one of the other spaces, and who has been selected as the policy recipient in a space in which a related feature having a higher degree of relevance to the evaluation index is set as an axis.

[0285] This makes it possible to appropriately select subjects for whom measures are likely to be effective even when a large number of related features are identified.

[0286] (7) In the policy planning support system described in (3) above, the processor selects the policy recipients based on the matching rate of subjects within a predetermined area in each of the spaces.

[0287] This makes it possible to appropriately select subjects for whom measures are likely to be effective even when a large number of related features are identified.

[0288] (8) In the policy planning support system described in (7) above, when the matching rate of subjects within the specified area in each of the spaces is higher than a specified standard, the processor outputs information about the subjects selected as the policy recipients in common in each of the spaces as information to support the policy planning.

[0289] This makes it possible to appropriately select subjects for whom measures are likely to be effective even when a large number of related features are identified.

[0290] (9) In the policy planning support system described in (7) above, when the matching rate of subjects within the specified area in each of the spaces is lower than a specified standard, the processor outputs information about the subjects selected as policy recipients in each of the spaces as information to support the policy planning.

[0291] This makes it possible to appropriately select subjects for whom measures are likely to be effective even when a large number of related features are identified.

[0292] (10) In the policy planning support system described in (7) above, the predetermined area in each of the spaces is at least one of the role model area and the policy adopter area.

[0293] This makes it possible to appropriately select subjects for whom measures are likely to be effective even when a large number of related features are identified.

[0294] (11) In the policy planning support system described in (3) above, the number of dimensions of each of the spaces is three or less.

[0295] This makes it possible to display the subject's features in a visually comparable manner even when a large number of related features are identified.

[0296] (12) A policy planning support system as described in (2) above, wherein the space is divided into a plurality of sections (for example, space example 607 shown in Figure 6), and the processor identifies at least one of the sections including the position where the related features of the role model are projected as the role model area, identifies at least one of the sections adjacent to the role model area as the policy recipient area, and selects the subject corresponding to the related features projected into the policy recipient area as the policy recipient.

[0297] This allows appropriate selection of subjects for whom the measures are likely to be effective.

[0298] (13) In the policy planning support system described in (2) above, the processor selects the subject person corresponding to the related feature projected at a position closer than a predetermined standard from the role model area as the policy recipient (for example, space example 606 shown in Figure 6).

[0299] This allows appropriate selection of subjects for whom the measures are likely to be effective.

[0300] (14) In the policy planning support system described in (2) above, the storage device further stores feature information (e.g., feature table 200) that associates each of the multiple features with a policy to be taken for the target person, and the processor outputs, as information to support the policy planning, the policy associated with the related feature set as an axis in the space onto which the related feature of the policy recipient is projected.

[0301] This will assist in planning measures that are expected to be effective in improving KPIs.

[0302] (15) In the policy planning support system described in (2) above, the storage device stores role model selection conditions (e.g., role model selection conditions 207) indicating conditions for selecting the role model from the subjects based on the features, and the processor projects the related features of each subject before and after implementing measures to improve evaluation indexes into space (e.g., example feature transitions 2412 and 2413 in Figure 24C, example feature transitions 2703 and 2704 in Figure 27A), and evaluates the effectiveness of the measures based on changes in the number of subjects who satisfy the role model selection conditions.

[0303] This allows the effectiveness of the measures to be properly evaluated.

[0304] (16) In the policy planning support system described in (2) above, the processor projects the related features of the multiple subjects into the space before and after implementing a policy to improve the evaluation index (for example, feature transition examples 2412 and 2413 in Figure 24C, and feature transition examples 2703 and 2704 in Figure 27A), and evaluates the effectiveness of the policy based on whether the position where the related features of the policy recipient are projected approaches the role model area.

[0305] This allows the effectiveness of the measures to be properly evaluated.

[0306] (17) In the policy planning support system described in (1) above, the processor calculates the relationship between the evaluation index and the feature based on at least one of the correlation between the evaluation index and the feature and the contribution of the feature in a model that infers the value of the evaluation index from the value of the feature.

[0307] This allows for an appropriate evaluation of the relationship between KPIs and feature quantities.

[0308] It should be noted that the present invention is not limited to the above-described embodiment, and includes various modifications. For example, the above-described embodiment has been described in detail to provide a better understanding of the present invention, and the present invention is not necessarily limited to an embodiment having all of the configurations described.

[0309] Furthermore, the above-described configurations, functions, processing units, processing means, etc. may be partially or entirely implemented in hardware, for example, by designing them as integrated circuits. The above-described configurations, functions, etc. may also be implemented in software, with a processor interpreting and executing a program that implements each function. Information such as the programs, tables, and files that implement each function can be stored in storage devices such as nonvolatile semiconductor memory, hard disk drives, and solid-state drives (SSDs), or in computer-readable, non-transitory data storage media such as IC cards, SD cards, and DVDs.

[0310] In addition, the control lines and information lines shown are those that are considered necessary for the explanation, and not all control lines and information lines in the product are necessarily shown. In reality, it can be considered that almost all components are interconnected. [Explanation of symbols]

[0311] 100 Policy Planning Support System 101 Data input / output unit 102 databases 103 KPI Selection Section 104 Feature Selection Unit 105 Mapping Department 106 Feature Projection Unit 107 Role Model Selection Department 108 Policy Applicant Selection Department 109 Spatial Integration Department 110 Policy Selection Department 111 Policy Effectiveness Verification Department 112 Visualization section 200 Feature Table 201 KPI information 202 Related Features 203 Related feature selection conditions 204 Target information 205 Spatial Information 206 Role Model Information 207 Role Model Selection Criteria 208 Policy applicator information 209 Selection criteria for policy recipients 210 Integrated Spatial Information 211 Integrated Spatial Conditions 212 Policy information 213 Transition information 214 Effectiveness Confirmation Indicator Information 215 Setting Information

Claims

1. A policy planning support system, a processor and a storage device, the storage device holds subject information including a plurality of feature amounts relating to each of a plurality of subjects; the plurality of feature amounts include at least one of a feature amount related to the behavior of the subject and a feature amount related to the attribute of the subject; The processor: identifying, as related feature quantities, two or more feature quantities related to the feature quantities designated as the evaluation indexes, among the plurality of feature quantities; projecting the related feature of each subject onto a space having axes of the two or more related feature; selecting, from the plurality of subjects, one or more subjects whose evaluation level based on the feature amount designated as the evaluation index satisfies a predetermined standard, as role models; A policy planning support system characterized by outputting information for displaying the space onto which the related features of multiple subjects, including the role model, are projected as information to support policy planning for improving the evaluation index.

2. 2. The policy planning support system according to claim 1, The processor: Identifying a region in the space that includes a position onto which the feature amount of the subject selected as the role model is projected as a role model region; Identifying an area within a predetermined range from the role model area as a policy recipient area; selecting one or more subjects corresponding to one or more feature amounts projected onto the policy recipient area as policy recipients; A policy planning support system characterized in that it outputs information about the policy applicant as information for supporting the policy planning.

3. 3. The policy planning support system according to claim 2, The processor: generating combinations of two or more of the related features; generating the space onto which the related feature of each subject is projected for each of two or more combinations of the related feature; A policy planning support system characterized by selecting the policy applicant in each of the spaces.

4. 4. The policy planning support system according to claim 3, A policy planning support system characterized in that the processor selects the policy applicant based on the degree of relevance between the related features set as axes in each of the spaces and the evaluation index.

5. 5. The policy planning support system according to claim 4, A policy planning support system characterized in that the processor outputs information about subjects selected as policy recipients in common across multiple spaces as information to support the policy planning.

6. 6. The policy planning support system according to claim 5, The processor outputs, as information to support the policy planning, information regarding subjects who have been selected as policy recipients in at least one of the multiple spaces and who have not been selected as policy recipients in at least one of the other spaces, in a space in which a related feature value having a higher degree of relevance to the evaluation index is set as an axis.

7. 4. The policy planning support system according to claim 3, A policy planning support system characterized in that the processor selects the policy recipients based on the matching rate of subjects within a specified area in each of the spaces.

8. The policy planning support system according to claim 7, A policy planning support system characterized in that, when the matching rate of subjects within the specified area in each of the spaces is higher than a specified standard, the processor outputs information regarding the subjects selected as policy recipients in common in each of the spaces as information to support the policy planning.

9. The policy planning support system according to claim 7, A policy planning support system characterized in that, when the matching rate of subjects within the specified area in each of the spaces is lower than a specified standard, the processor outputs information regarding the subjects selected as policy recipients in each of the spaces as information to support the policy planning.

10. The policy planning support system according to claim 7, A policy planning support system, characterized in that the predetermined area in each of the spaces is at least one of the role model area and the policy adopter area.

11. 4. The policy planning support system according to claim 3, A policy planning support system characterized in that the number of dimensions of each of the spaces is three or less.

12. 3. The policy planning support system according to claim 2, The space is divided into a plurality of compartments, The processor: identifying at least one of the sections including a position where the related feature amount of the role model is projected as the role model area; Identifying at least one of the sections adjacent to the role model area as the policy recipient area; A policy planning support system characterized by selecting, as the policy recipient, the subject person corresponding to the related feature projected onto the policy recipient area.

13. 3. The policy planning support system according to claim 2, A policy planning support system characterized in that the processor selects the subject person corresponding to the related feature projected at a position closer than a predetermined standard from the role model area as the policy recipient.

14. 3. The policy planning support system according to claim 2, the storage device further stores feature amount information that associates each of the plurality of feature amounts with a measure to be taken for the subject; The processor outputs the policy associated with the related feature set as an axis in the space onto which the related feature of the policy recipient is projected as information to support the policy planning.

15. 3. The policy planning support system according to claim 2, the storage device holds role model selection conditions indicating conditions for selecting the role model from the subject based on the feature amount; The processor: The relevant features of each subject before and after implementing measures to improve the evaluation index are projected onto the space. A policy planning support system characterized by evaluating the effectiveness of the policy based on changes in the number of subjects who satisfy role model selection conditions.

16. 3. The policy planning support system according to claim 2, The processor: projecting the related feature amounts of the plurality of subjects before and after implementing measures to improve the evaluation index into the space; A policy planning support system characterized in that the effectiveness of the policy is evaluated based on whether the position where the relevant feature amount of the policy recipient is projected approaches the role model area.

17. 2. The policy planning support system according to claim 1, The processor calculates the relationship between the evaluation index and the feature based on at least one of the correlation between the evaluation index and the feature and the contribution of the feature in a model that infers the value of the evaluation index from the value of the feature.

18. A policy planning support method executed by a computer system, comprising: the computer system includes a processor and a storage device; the storage device holds subject information including a plurality of feature amounts relating to each of a plurality of subjects; the plurality of feature amounts include at least one of a feature amount related to the behavior of the subject and a feature amount related to the attribute of the subject; The policy planning support method includes: a step by the processor of identifying, as related feature quantities, two or more feature quantities related to a feature quantity designated as an evaluation index, among the plurality of feature quantities; a step of the processor projecting the related feature of each of the subjects onto a space having axes each of the two or more related feature; a step by the processor selecting, from the plurality of subjects, one or more subjects whose evaluation level based on the feature amount specified as the evaluation index satisfies a predetermined standard as a role model; and a step in which the processor outputs information for displaying the space onto which the related features of the plurality of subjects, including the role model, are projected, as information for supporting the formulation of measures to improve the evaluation index.

Images