JP2002279148A - Data analysis program, data analysis method using computer system and data analysis system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately divide analysis object data collected at a plurality of stages into groups. SOLUTION: According to this data analysis program 1, in order to classify newly collected analysis object data, a computer realizes a determining function of determining the compatibility between an already generated group and newly collected analysis object data according to the previously collected analysis object data, a first classifying function 3 of classifying the analysis object data determined to be compatible by the determining function 4, and a second classifying function 5 of generating a new group according to the analysis object data determined to be incompatible by the determining function 4 and classifying the analysis object data determined to be incompatible into the new group.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a data analysis program for classifying and analyzing collected data, a data analysis method by a computer system, and a data analysis system.

[0002]

2. Description of the Related Art An analyst who analyzes a result of a marketing survey collects data such as consumer information and product information (hereinafter referred to as "analysis target data") by, for example, a questionnaire survey and performs analysis. . Various analysis methods have been proposed, for example, using multivariate analysis. The analyst uses an analysis method suitable for the content of the data to be analyzed and the content of the analysis.

At the time of analysis, clustering, classification trees,
The data to be analyzed may be classified using discriminant analysis or the like. By performing such a classification, for example, some groups are generated for consumers and commodities.

FIG. 10 is a diagram illustrating a conventional classification method based on the result of a marketing survey.

[0005] In this example, groups 1G and 2G are generated based on the collected data to be analyzed. Boundary A
Is determined as a reference, whether it belongs to the group 1G or the group 2G. For example, the boundary A is determined by a distance from a cluster center in clustering, a discriminant function in discriminant analysis, a classification tree, and the like.

[0006] In the conventional classification in the marketing survey, when new analysis target data is collected, even if the newly collected classification target data has a different property from the previously collected classification target data, Based on the boundary A, it is determined whether it is a group 1G or a group 2G.

[0007] The analyst performs an action according to the analysis result (grouping result, classification result) with reference to the analysis result.

[0008]

As described above, in the conventional classification of the results of marketing research, when a new group of data to be analyzed is collected after a group of data to be analyzed is collected, the newly collected analysis data is analyzed. The target data is assigned to one of the previously generated groups.

In this case, the newly collected analysis target data is assigned to the generated group even if it contains unique characteristics different from the previously collected analysis target data.

Therefore, it is difficult to perform accurate grouping in the conventional classification in marketing research. Further, the characteristics of the newly collected analysis target data do not affect the classification result, and thus are not effectively used for analysis.

Also, after analyzing the results of the marketing survey, various actions may be taken in accordance with the results of the analysis (grouping results). Attempts have not been fully explored.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and has a data analysis program for performing highly accurate grouping and analysis using the grouping result, a data analysis method using a computer system, and a data analysis. The purpose is to provide a system.

[0013]

Means taken to realize the present invention will be described below.

According to a first aspect of the present invention, in order to classify newly collected analysis target data, a computer includes a group generated based on previously collected analysis target data and a newly collected analysis target data. And a first classification function for classifying the analysis target data determined to be suitable by the determination function into groups that have already been generated, based on the analysis target data determined to be nonconforming by the determination function A new group is generated, and the analysis target data determined to be non-conforming is classified into this new group.
It is a data analysis program for realizing the classification function.

According to a second aspect of the present invention, in the data analysis program according to the first aspect of the present invention, a computer is provided with a feature extracting function for extracting a feature of analysis target data included in each generated group. The program has a program for realizing it, and the determination function compares the generated group with the newly collected analysis target data by comparing the feature extracted by the feature extraction function with the newly collected analysis target data. The compatibility between the two.

In the first and second aspects of the present invention, a new group is generated when the newly collected data to be analyzed is not suitable for any of the already generated groups.

Therefore, it is possible to obtain an analysis result reflecting the characteristics of the newly collected analysis target data.

According to a third aspect of the present invention, a computer is provided with a feature extracting function for extracting features of data to be analyzed included in a group generated based on collected data to be analyzed, and features extracted by the feature extracting function. A data analysis program for realizing a position information extracting function of extracting position information of data to be analyzed in a group based on a difference between the data to be analyzed and data to be analyzed included in the group.

According to the third aspect of the present invention, the characteristics of the group can be extracted and the position of the classified data to be analyzed within the group can be grasped. I can clarify.

According to a fourth aspect of the present invention, there is provided a computer having a feature extracting function and a feature for extracting, for each of a plurality of groups generated based on collected analysis target data, the characteristics of the analysis target data included in the group. This is a data analysis program for realizing a function of displaying the characteristics of a plurality of groups extracted by the extraction function in a comparable form.

In the fourth aspect, the characteristics between groups can be compared. Thereby, it is possible to grasp how to change the characteristics of the analysis target data in order for the analysis target data classified into a certain group to be classified into another certain group.

Therefore, the action after the analysis can be easily determined.

The data analysis method implemented by the program according to the first to fourth inventions may be an object of the invention.

Further, a data analysis system that realizes the functions realized by the programs according to the first to fourth inventions may be an object of the invention.

[0025]

Embodiments of the present invention will be described below with reference to the drawings. In the drawings described below, the same portions are denoted by the same reference numerals, and the description thereof will be omitted. Only different portions will be described in detail.

(First Embodiment) The concept of this embodiment will be described.

When the data to be analyzed in marketing research and analysis is classified, some groups are generated.

Normally, the newly collected analysis target data is classified into any one of the groups generated for the previously collected analysis target data.

However, when the previously classified analysis target data is compared with the newly collected analysis target data,
The newly collected analysis target data may not retain the characteristics of the previously generated group so strongly.

In this embodiment, in such a case, a new group is generated based on newly collected data to be analyzed.

Even if the same conditions are set and the data to be analyzed is collected, if the time to collect the data to be analyzed is different, the state of a person or an object also changes due to a change in the surrounding environment. If newly collected data to be analyzed can be classified into the most appropriate group without being constrained by existing classification results, the classification accuracy of the data to be analyzed will improve, and more accurate actions will be taken for the group. It becomes possible.

The data to be analyzed included in the same group is considered to have features common to the group.

However, by slightly changing the characteristics of the data to be analyzed included in the group, some data to be analyzed belonging to the group can be processed so as to be included in another group.

For example, if there is a difference between a certain group obtained as a result of the classification and the group targeted by the analyst, what kind of action should be taken to make the certain group the group targeted by the analyst. It can be determined that it can be changed.

If the characteristics of the group can be examined and the data to be analyzed can be moved between the groups, it is possible to formulate and execute an accurate marketing strategy.

FIG. 1 is a block diagram illustrating the configuration of a data analysis program according to the present embodiment for realizing the above concept. FIG. 1 shows a data analysis program 1
Is a function realized on the computer which has read from the recording medium 2.

The data analysis program 1 according to the present embodiment reads data to be analyzed collected in a plurality of stages and performs classification for analysis.

The data analysis program 1 implements a classification function 3, a judgment function 4, a new classification function 5, a feature extraction function 6, a position information extraction function 7, and a comparison function 8 on a computer or a computer system.

The data analysis program 1 accesses a database 9 for registering previously collected analysis target data, a new database 10 for registering newly collected analysis target data, and a classification database 11.

The classification function 3 receives and classifies the data to be analyzed which has already been collected, generates a group, and registers the result in the classification database 11. In this example, a group 1G and a group 2G are generated from the analysis target data that has been collected first.

The classification function 3 receives new analysis target data determined to be included in the generated group from the determination function 4 and classifies the received new analysis target data into any of the generated groups. Judge whether to do it, and register it in an appropriate group.

The determination function 4 determines whether the newly collected analysis target data matches any of the generated groups. The determination of the suitability is performed, for example, by comparing newly collected analysis target data with the features of each group extracted by the feature extraction function 6.

If the determination function 4 determines that the newly collected analysis target data matches one of the generated groups, the determination function 4 provides the newly collected analysis target data to the classification function 3. I do.

On the other hand, if it is determined that the data does not match any of the generated groups, the determination function 4 provides the newly collected analysis target data to the new classification function 5.

The new classification function 5 receives analysis target data determined not to be included in the generated group from the determination function 4, generates a new group based on the received analysis target data, and classifies the result. Register in the database 11. In this example, the group 3G is generated from newly collected analysis target data.

The feature extracting function 6 extracts features of each group. The features are extracted by, for example, totaling one or both of the answer content of the analysis target data and the respondent information included in each group. For example, when the average age of the respondents of the analysis target data belonging to the group 1G is 30 years, this is a feature of the group 1G. In addition, Group 1
The gender of the respondents of the analysis target data belonging to G is 9 for males
In the case of 0%, it is a feature of the group 1G that 90% of men are present.

The position information extracting function 7 extracts position information indicating the position of the data to be analyzed belonging to the group in each group, and provides it to the analyst.

The comparison function 8 displays the characteristics of a plurality of groups in a form that can be compared.

A data analysis method using the data analysis program 1 having the above configuration will be described.

FIG. 2 is a flowchart illustrating a data classification method by the data analysis program 1 according to the present embodiment.

First, group generation and classification based on collected data are executed by the classification function 3 (S1).

Next, the newly collected analysis target data is received by the determination function 4 (S2).

Next, the determination function 4 determines the compatibility between the generated group and the newly collected analysis target data.
(S3).

If it is determined that the data is suitable in the above determination, the newly collected analysis target data is classified by the classification function 3 into a group that has already been generated (S4). A new group is generated by the new classification function 5 based on the newly collected analysis target data, and the newly collected analysis target data is classified into this new group (S5).

FIG. 3 is a flowchart illustrating a method of positioning and outputting analysis target data classified by the data analysis program 1 according to the present embodiment.

First, the feature of each group is represented by a feature extraction function 6.
(T1).

Next, the specification of the analysis target data whose position is desired to be grasped is received from the user (T2).

Next, based on the difference between the specified analysis target data and the characteristics of the group, the position information of the analysis target data in this group is extracted by the position information extracting function 7 and provided to the user ( T3).

FIG. 4 is a flowchart illustrating a method of comparing groups generated by the data analysis program 1 according to the present embodiment.

First, a user specifies a group to be compared (U1).

Next, the features of each group extracted by the feature extraction function 6 are extracted by the comparison function 7 (U
2).

Then, the extracted features of each group are displayed in a form that can be compared by the comparison function 7 (U3).

Hereinafter, a consumer questionnaire survey in the marketing field using the data analysis program 1 according to the present embodiment will be described.

The survey assumed here relates to a certain product Z, and it is assumed that answers to question items regarding “price setting”, “product image”, “function”, etc. of this product Z have been obtained for each respondent. Assume.

The appropriate classification of the data to be analyzed will be described with reference to FIG.

The distribution 12 shows a result of analyzing answers to a plurality of questions for each respondent and assigning a score.

In the figure, one of “•” indicates one respondent. Based on this result, it is assumed that grouping is performed using a technique such as clustering, and a result like a distribution 13 is obtained. In the distribution 13, the respondents are classified into two groups 1G and 2G.

For example, assume that group 1G is a group of respondents who have already purchased product Z, and group 2G is a group of respondents who have not yet purchased product Z.

It is assumed that a short time after this questionnaire survey, a questionnaire was taken to another consumer with the same question item, and that data to be analyzed was obtained from a new respondent “×”.

In the normal analysis, the answer of the respondent "x" is analyzed to determine which of the generated groups 1G and 2G is included. This state is distribution 14. In this example, the new respondent "x" is determined to be included in 2G rather than group 1G, indicating that the range of group 2G has been expanded.

However, at first glance, the respondent “×” is compared with the other respondents in the group 2G,
It appears that the claim that it should be included in

Therefore, in the present embodiment, it is first determined whether or not the respondent "X" may be included in the groups 1G and 2G. To generate a new group 3G. Distribution 15 shows this state.

In this example, the newly generated group 3G
In the determination of suitability that the group was interpreted as being in the group “Product Z has not been purchased yet, there is a possibility that purchase will be considered in the future”, for example, respondents “·” of groups 1G and 2G Is measured (distance between individual data), and how the response "*" changes when the respondent enters. Alternatively, the center of gravity of the group is determined, and the degree to which the respondent "x" is far from the center of gravity is determined, and the determination is made.

FIG. 6 illustrates an example in which the determination is made based on the distance from the center of gravity. If a threshold value is determined in advance, and the distance between the respondent "x" and the center of gravity of the generated group exceeds this threshold, it is determined that "the respondent" x "should not be included in the generated group" Is done.

The position information extracting function 7 realized by the data analysis program 1 displays the distance information between the data to be analyzed included in a certain group and the center of gravity as position information to the analyst.

It is assumed that the classification is performed according to the normal classification method, and that the respondent "x" is determined to be the generated group 2G (= not purchased).

In this case, there is a difference in the planning of the next marketing strategy between the case where it is determined that the group is 2G and the case where it is determined that it is another group 3G (= unpurchased which may be purchased).

If the respondent "x" can be judged to be in the group 3G, it is possible to make a strategy such as sending a product catalog or strongly appealing the privilege when purchasing.

Further, by repeatedly performing the same questionnaire survey, changes in grouping can be grasped in time series. As a result, a long-term marketing strategy can be established.

Next, a method of changing a group will be described with reference to FIG.

The classified group of the data to be analyzed is
Each has certain characteristics. Group 1G = "A group of people who have purchased the product Z", Group 2G = "A group of people who have not yet purchased the product Z", Group 3G = "A group of people who have not purchased the product Z but are likely to buy it" Is the feature.

These characteristics, when viewed in more detail, consist of several factors. For example, to form a group 2G of persons who have not yet purchased the product Z is “product Z
Function (I don't understand much), age (approximately 30
And the type of job (there is a strong tendency for brain labor), and so on. "

Here, the following notation is used to simply represent the relationship between a feature and its factor. For example, three groups 1G, 2G, there is a 3G, each feature is that consists factor _{r 1, r 2, r 3} , r 4. FIG.
Shows a case where the state of the characteristic factor of each group can be expressed in a form such as a two-dimensional graph.

If the degree of influence of a factor is represented by a numerical value arranged before the factor, for example, the feature of group 1G is R (2r ₁ , r ₂ , 3r ₃ , r ₄ ),
Features of G is _{R (r 1, r 2,} 2r 3, r 4), wherein the groups 3G are represented as the _{R (2r 1, r 2,} 2r 3, r 4).

[0085] When attention is paid to the differences between the groups, consumers belonging to the group 3G is to strengthen the factors r ₃ Group 1
It can be clearly understood that it is possible to belong to G.

In other words, for example, if the respondent who has not purchased the product Z at this time but has a high possibility of purchasing the product Z can learn more about the function of the product Z, he or she can shift to the respondent who purchases the product Z. I can judge.

[0087] In addition, focusing on the differences between the groups, differences between groups 2G and Group 3G, it can be determined that is that there is a difference in the intensity of the influence of factors r _1. Therefore,
For example, if the factors r ₁ is assumed to be age, by performing the promotion of products Z for the elderly, commodity Z "Product Z not yet purchased" respondent is "currently there is a high possibility that but not yet purchased to buy "We can judge that we will shift to respondents.

If such post-analysis action can be taken, “develop an advertisement that introduces the function of the product Z in detail in each medium” or “medium that the respondents of the group 3G often see. It is possible to plan and execute a lean and effective strategy such as "development of advertisements in a narrow manner" and to clarify guidelines for how to develop a marketing strategy.

FIG. 8 shows a comparison result of the characteristics between the groups displayed by the comparison function 8 by the data analysis program 1 according to the present embodiment. This comparison result can be used as a reference when attempting to change the group as described above.

The effects obtained by using the data analysis program 1 according to the present embodiment as described above will be summarized.

First, when the data analysis program 1 is used, if the generated group does not accurately reflect the characteristics of the newly collected analysis target data, a new group is generated based on the newly collected analysis target data. Is generated. This enables accurate grouping.

Next, by using the data analysis program 1, it is possible to grasp the position of the data to be analyzed within each group and to clarify how strongly the data to be analyzed has inherited the characteristics of the group. .

Next, when the data analysis program 1 is used, the analyst can grasp the factors characterizing each group obtained as a result of classifying the data to be analyzed, and can clearly recognize the difference.

[0094] Then, the analyst can easily grasp what action should be performed to move certain analysis target data to another group.

The functions realized by the data analysis program 1 according to the present embodiment may be changed in arrangement as long as similar functions can be realized, or the functions may be freely combined.

The data analysis program 1 described in the present embodiment includes a recording medium 2 such as a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), an optical disk (CD-ROM, DVD, etc.), a semiconductor memory, etc. And can be applied to a computer. Further, this program is transmitted through a communication medium, and is applicable to computers and computer systems.

The computer reads the data analysis program 1 recorded on the recording medium 2 from the recording medium 2 and the operation is controlled by the program 1 so that
Implement the functions described above.

(Second Embodiment) In the present embodiment, the use of the data analysis program 1 according to the first embodiment will be described.

FIG. 9 shows a service executed by the data analysis program 1 according to the first embodiment as an ASP.
It is a block diagram which illustrates the form provided by (application service provider).

The user (analyzer) 16 has his / her own terminal 17
By using the data analysis program 1 managed by the ASP 19 via the network 18, the analysis of the analysis target data can be performed efficiently and easily.

Further, by receiving the service provided by the ASP 18, the user 16 can use the data analysis service more efficiently in terms of maintenance and operation than when the data analysis program 1 is managed by himself.

The ASP 18 provides a data analysis service to the user 16 and can receive compensation from the user 16.

[0103]

As described in detail above, in the present invention,
If the newly collected data to be analyzed does not match the generated group, a new group is generated and classified into the new group.

Accordingly, accurate grouping can be performed, and the accuracy of analysis can be improved.

In the present invention, the characteristics of each group are extracted, and the position of the data to be analyzed included in the group or the difference in the characteristics between the groups can be easily grasped.

Therefore, it is possible to plan and execute an effective marketing strategy. For example, by narrowing down the target by advertising development or the like, the cost for marketing can be suppressed.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating the configuration of a data analysis program according to a first embodiment of the present invention.

FIG. 2 is a flowchart illustrating a data classification method by the data analysis program according to the embodiment;

FIG. 3 is a flowchart illustrating a method of positioning and outputting analysis target data classified by the data analysis program according to the embodiment;

FIG. 4 is an exemplary flowchart illustrating a comparison method between groups generated by the data analysis program according to the embodiment.

FIG. 5 is a diagram illustrating a classification state of data to be analyzed;

FIG. 6 is a diagram exemplifying a method of determining the suitability of a group and analysis target data;

FIG. 7 is a diagram illustrating an example of changing a group.

FIG. 8 is a diagram illustrating a comparison result between groups.

FIG. 9 is a block diagram illustrating a form in which an ASP provides a service performed by a data analysis program.

FIG. 10 is a view exemplifying a conventional classification method for a result of conducting a marketing survey.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Data analysis program 2 ... Recording medium 3 ... Classification function 4 ... Judgment function 5 ... New classification function 6 ... Feature extraction function 7 ... Location information extraction function 8 ... Comparison function

Claims (10)

[Claims] 1. A computer for classifying newly collected data to be analyzed, wherein a computer matches a group generated based on previously collected data to be analyzed with newly collected data to be analyzed. A first classification function of classifying the analysis target data determined to be suitable by the determination function into the generated groups; a determination function determining the non-conformity by the determination function based on the analysis target data A data analysis program for generating a new group and implementing a second classification function of classifying the analysis target data determined to be nonconforming into the new group. 2. The data analysis program according to claim 1, further comprising: a program for causing a computer to implement a feature extraction function for extracting a feature of analysis target data included in each of the generated groups. The determination function compares the feature extracted by the feature extraction function with the newly collected analysis target data to determine the compatibility between the generated group and the newly collected analysis target data. A data analysis program characterized by determining. 3. A feature extraction function for extracting features of analysis target data included in a group generated based on collected analysis target data, a feature extracted by the feature extraction function and the group A data analysis program for realizing a position information extracting function of extracting position information of the analysis target data in the group based on a difference from the analysis target data included in the data. 4. A feature extracting function for extracting, for each of a plurality of groups generated based on the collected analysis target data, a feature of the analysis target data included in the group; A data analysis program for realizing a comparison function of displaying extracted features of a plurality of groups in a comparable form. 5. A method for analyzing data by a computer system, comprising: a receiving step of receiving newly collected analysis target data; a group generated based on previously collected analysis target data; and a newly collected analysis target. A determination step of determining compatibility between the data and, if determined to be suitable by the determination step, classify the newly collected analysis target data into the generated group, and determine that the data is incompatible by the determination step. A classification step of generating a new group based on the newly collected analysis target data when the determination is made, and classifying the non-conforming analysis target data into the new group. 6. A method for analyzing data by a computer system, comprising: a step of extracting characteristics of data to be analyzed included in a group generated based on the collected data to be analyzed; Extracting the position information of the data to be analyzed in the group based on the difference from the data to be analyzed included in the data analysis method. 7. A method of analyzing data by a computer system, wherein the step of receiving designation of a plurality of groups generated based on the collected analysis target data from a user is included in the plurality of groups designated by the user. A data analysis method comprising: a step of extracting characteristics of data to be analyzed; and a step of displaying the extracted characteristics of a plurality of groups in a comparable form. 8. A system for analyzing data to be analyzed, a determination unit for determining compatibility between a group that has been generated based on data that has been previously collected and data that has been newly collected. A first grouping unit that classifies the analysis target data determined to be suitable by the determination unit into the generated group; and generates a new group based on the analysis target data determined to be nonconforming by the determination unit. And a second classifying means for classifying the analysis target data determined to be non-conforming into the new group. 9. A system for analyzing data to be analyzed, wherein: a feature extracting unit for extracting features of the data to be analyzed included in a group generated based on the collected data to be analyzed; Data analysis, comprising: position information extracting means for extracting position information of data to be analyzed in the group based on a difference between the extracted feature and data to be analyzed included in the group. system. 10. A system for analyzing data to be analyzed, wherein, for each of a plurality of groups generated based on the collected data to be analyzed, feature extracting means for extracting characteristics of the data to be analyzed included in the group. And a means for displaying the features of a plurality of groups extracted by the feature extracting means in a comparable form.