JPH06162063A - Data processing system - Google Patents

Abstract

(57) [Summary] [Purpose] Obtaining all items that have a causal relationship with each other by freely and efficiently synthesizing multiple sets of simple correlations representing causal relationships between data items according to the operator's instructions. . [Structure] Data before edit is input for all items or specified items from the normal purpose-use database as is (101), and then a simple correlation table for all items is output to an output device (102). Enter the levels in order of causality (103), and correlate between items (causality)
The lowest possible correlation coefficient is designated (104), the item array order is designated as ascending or descending correlation coefficient (105), and the simple correlation table is searched based on the designated items. Then, a plurality of sets of simple correlations are comprehensively aggregated and a correlation relation table showing multiple correlations between these items is created (106). The aggregated data is stored in the storage device (107).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data processing method used as a part of performance evaluation of a company such as a trading company or a financial institution, and in particular, it relates correlation between various business evaluation items and business performance items of a company. It relates to a data processing method that searches for the cause of sales performance by investigating.

[0002]

2. Description of the Related Art In recent years, financial institutions and the like have been required to clarify the relationship between performance evaluation items and sales performance items as the needs for performance evaluation increase. For example,
It is necessary to extract various data items related to causes and effects of savings, lending, profits, interest rate fluctuations, etc., examine the relationships between these items, and find the cause item for the specified item.

In a trading company or the like, for example, a factor for purchasing a hot-selling product or a dying-selling product and its cause are found,
It has become necessary to perform data output processing so that items that cause changes in the purchasing power of these products can be easily found.

As a data processing method of this kind, conventionally, for example, Haruo Yanai, Shuichi Iwatsubo, "Science to Challenge Complexity (Introduction to Multivariate Analysis)" (published on August 20, 1985, Kodansha), page 31- As described on page 49 (Reference 1), a method is known in which the correlation between items is calculated by calculating the correlation coefficient of the single correlation between the items.

Further, for example, "Hitachi Creative Workstation 2050 Total Calculation Library HI
39 of "STAT 1" (issued in 2nd edition, August 1989)
As described in pages 2 to 42 (Document 2), a method of outputting a correlation table (a cross-correlation coefficient matrix of items) according to items designated by a user is known.

[0006]

[Problems to be Solved by the Invention] In the method of evaluating the performance of a company by examining the correlations between the above-mentioned sales evaluation items and sales performance items, the correlation analysis method required in the future is the cause between the items. It is to clarify the causal relationship between the results and the results, to find all the causes for a certain result, and to find further causes for these causes, and to elicit all the causes without fail. As one of the methods for this purpose, a method of extracting all direct or indirect causes for a certain result and classifying (defining) the causes according to types can be considered. However, at present, there is only a single correlation table as in the above-mentioned literature, and each item and its correlation coefficient are manually extracted by using this correlation table, and the single correlation coefficient of one item is calculated. However, there is a problem that it is difficult to find all the correlations with respect to a certain item in a short time and efficiently without fail.

Therefore, an object of the present invention is to overcome the above-mentioned problems of the prior art and to easily create a correlation correlation table or correlation correlation diagram of multiple correlations based on the conventional correlation table based on a single correlation. An object of the present invention is to provide a data processing method capable of efficiently finding all the causal relations between the items of data including the items in a short time and efficiently.

[0008]

In order to achieve the above object, the data processing system of the present invention comprises means for inputting all items or items designated by a user from a purpose-specific database into a management table and a correlation analysis method. A correlation table storing a single correlation coefficient between each item, and a correlation relation table having a multiple correlation between the plurality of items by searching the correlation table based on a plurality of items arbitrarily designated by the user, or And means for creating a correlation diagram.

Here, a plurality of items arbitrarily designated by the user are sequentially classified into levels according to the mutual causal relationship (relationship between the result item and the cause item) between the items, and the correlation relation table or the correlation relation diagram. Shows that each item is divided and arranged corresponding to the level division. That is, a result item, one or more first cause items that cause the result item, and one or more second cause items that cause each of the first cause items (the same applies below). The items are sequentially classified into levels, and each item is arranged in the correlation-related table or the correlation-related diagram according to the level.

In addition, based on the means for designating the correlation coefficient value of a predetermined value or more by the user and the designation of the correlation coefficient value,
It is provided with means for searching for an item having a correlation coefficient value equal to or more than the predetermined value among a plurality of items arbitrarily specified by the user and outputting the item to the correlation coefficient table or the correlation coefficient diagram.

Further, means for inputting array designation information for designating whether the level-divided items are arranged in the ascending order or the descending order of levels, and the correlation coefficient table or the correlation coefficient according to the array designating information. Means for arranging the items in the figure may also be provided.

[0012]

The operation based on the above configuration will be described.

According to the present invention, the correlation coefficient of the single correlation between the respective items is obtained by the conventional correlation analysis method and stored in the correlation table, and the single correlation is calculated based on the plurality of items arbitrarily designated by the user. By searching the correlation table of the correlation, a correlation table or a correlation diagram having a multiple correlation between the plurality of items specified by the user is created, so that the user can directly or indirectly select a certain item. You can find all the cause items of all without fail. Further, a plurality of single correlation sets can be freely and efficiently combined from the correlation table in accordance with the instruction of the operator, and all the mutually related items can be surely found. In particular, a plurality of items arbitrarily designated by the user are classified into related levels, and a correlation related table or a correlation related diagram is automatically created by referring to the correlation table and searching based on the level-divided file. Therefore, this kind of totaling process (editing process) can be carried out strictly by level, omitting troublesome manual work, and efficiently performed in a short time without omission.
The accuracy of counting is improved.

Further, since the user can freely set and input the range of the correlation coefficient used for the totalization, for example, the minimum value, the totalization can be carried out freely.

The items in the output correlation table or correlation diagram can be arranged in either ascending order or descending order according to the user's designation.

[0016]

Embodiments of the present invention will be described below in detail with reference to the drawings.

1 and 2 are process flow charts showing an embodiment of the present invention. FIG. 3 is a device configuration diagram for realizing the present embodiment.

First, the apparatus configuration diagram shown in FIG. 3 will be described. The device configuration is such that the computer 301 includes an output device 305 such as a display, a storage device 302 and an input device 3.
03, 304 are connected. The storage device 302 is not limited to a magnetic storage device, but may be an optical disk or a semiconductor memory. In short, it is sufficient to have a sufficient capacity for executing the program and storing the data file.
In FIG. 3, a keyboard 303 and a mouse 3 are used as input devices.
Although two of 04 are shown, other input devices may be used, they may be used in combination, or any one of them may be used. As those skilled in the art can infer from the above description,
The apparatus configuration for implementing the present invention may be what is generally called a computer or a computer system. In particular,
It may be a large computer, workstation or personal computer. Further, in order to carry out the present invention,
It is needless to say that even the manufactured dedicated device may satisfy the functions of the peripheral devices as described above.

An outline of the operation in this embodiment will be described according to the processing flow of FIG. 1 with reference to FIG.

The unedited data collected in units of at least one set is used by the computer 3 using the input device 303.
01 is input (step 101). As shown in FIG. 4A, which will be described later, the pre-edit data is data obtained by directly inputting data for all items or items designated by a customer from a database for each purpose of use by a normal user. It consists of internal data such as and external data. Further, for example, the pre-edition data may be absolute value data (positive number) or percentage data, and may include character code data as well as numerical data. Next, a correlation table as shown in FIG. 4B described later is input using the input device 303 and output to the output device 305 (step 102). The correlation table itself can be obtained by the conventional single correlation analysis method as shown in Documents 1 and 2. In this embodiment, the correlation coefficient of the single correlation between the two items is obtained in advance by using this conventional correlation analysis method for all the items in the above-mentioned purpose-specific database and output to the output device. In FIG. 4B, the item name is shown by an actual name, but it may be shown by a number code as in Reference 2. Next, an item for each related level as shown in FIG. 5 described later is input using the input device 303 (step 103). Next, as shown in FIG. 6, which will be described later, a numerical range of the correlation coefficient that should be recognized as having correlation is designated and input using the input device 303 (step 104). Further, as shown in FIG. 6, the correlation coefficient increase / decrease specifying whether the correlated items are arranged in the ascending order or the descending order of the correlation coefficient.
The descending sort designation data is input using the input device 303 (step 105). The items are totaled in a correlation relation table (FIG. 7 described later) (step 106). The tabulated data is stored in the storage device 302 (step 107).

Note that once the related level-specific items are input, they can be repeatedly used when collecting different data. In that case, step 103 can be omitted in the flow of FIG.

Next, a first specific example of the data processing method based on the above embodiment will be described. Here, each processing step of FIG. 1 will be described in more detail with reference to FIGS. Figure 4
(A) is an example of a file format in which data before editing is stored, and as described above, is data in a database stored for each purpose of use by a normal user. Figure 4 (b)
Is an example in which the correlation coefficient of a single correlation is obtained in advance by the conventional method of Documents 1 and 2 and the correlation analysis result is output to the correlation table as described above. In FIG. 4B, the item names are shown as actual names, and all numerical values are correlation coefficients. FIG. 5 is a diagram showing an example of inputting items for each related level as described above, in which items (cause items) that are likely to be closely related to a certain result item by a user (customer, operator, etc.) are sequentially classified into levels. It was defined by. In FIG.
The relationships between level 1 and level 2, level 2 and level 3, level 3 and level 4, ... Are relationships between result items and their cause items. In other words, level 1 is an item (result item) that the user wants to investigate the cause, level 2 is an item causing level 1, level 3 is an item further causing each item of level 2, and so on. is there. First, the user arbitrarily selects an item (result item) whose cause is to be known and designates (defines) it as level 1. Next, the user thinks that items that are closely related to level 1 (cause items)
Are arbitrarily selected and designated (defined) as level 2. Next, each level 2 is used as a result item, and some of the terms (cause items) that cause it are arbitrarily selected and designated (defined) as level 3.

When defining these, FIG.
The user may arbitrarily select the item name from the unedited data 401 in (a), or the user may select the item name as he / she thinks of, and if necessary, the unedited data in FIG. 4 (a). The missing item name may be added to each level and defined by referring to the item name 401. Similarly, FIG. 5 is obtained.

As an alternative to the creation of FIG. 5, when the user first selects a desired result item from the pre-edit data items etc., which he / she wants to know the cause, and designates this as the level 1, it is automatically selected as shown in FIG. By referring to the correlation table 401 in (b), items having a slight correlation with the items of level 1 are extracted and displayed as candidates for level 2, and the user looks at this and considers that the causal relationship is not reversed. Level 2 items are defined by leaving only those that are judged to be truly related. Next, when each level 2 is designated (defined), level 3 candidates are automatically taken out, and when the levels are defined in the same manner, FIG. 5 is obtained. In FIG. 5, the “number of payees” means “the number of payees”.

FIG. 6 is a diagram showing an example of the processing content designation screen in tabulating the pre-edition data as described above. In editing, the user (customer or operator) selects and designates a highly correlated item. In addition, it is specified whether the order of arrangement is from low correlation or high correlation (ascending / descending sort). The correlation coefficient designation 1 is standard (0.
5 or more), 2 can be designated as an arbitrary value or more (designation of the minimum value). The designation in FIG. 6 can be arbitrarily made by the user. FIG. 7 is a diagram showing an example of a file format for storing data after totalization (details will be described later).

<Step 101> As data before editing, as shown in the “data before editing” of FIG. 4A, the performance evaluation items or sales performance items of the branch office of the regional bank (business net profit amount, personal liquidity) Deposit, pension recipient (other party to pension)
(Number, etc.) and numerical data are input to the management table from the keyboard 303 and stored in the form of a file 401.

<Step 102> Next, the result of the correlation analysis between the above items is output to the correlation table 402 as shown in the “correlation table” of FIG. 4B and stored in the storage device 302.

<Step 103> Next, as shown in “Relevance level” of FIG. 5, a file 501 in which items corresponding to each relevance level are entered is input from the keyboard 303 and stored in the storage device 302.

As described above, the level 1 corresponds to the result item, the level 2 corresponds to the cause item which is considered to have a strong correlation with the level 1 result item, and the level 3 corresponds to the level 2 (result item). Causes the items that seem to be strong, and similarly, from level 3 onwards, items that seem to have a strong correlation with the item immediately above correspond. Multiple items may correspond to the same level. Specify the causal relationship between items so that they do not reverse.

<Step 104> As a coefficient recognized as having correlation, a correlation coefficient is designated and input from the keyboard 303 on a processing content designation screen as shown in the screen 601 of FIG. 6 and stored in the storage device 302. For example, the correlation coefficient 0.
When extracting 5 or more, enter a correlation coefficient of 0.5.

<Step 105> The correlation coefficient ascending / descending sort designation data is designated and input from the keyboard 303 by the processing content designation screen as shown in the screen 601 of FIG. Here, in the items equal to or higher than the correlation coefficient input in step 104, the correlation coefficient is sorted in ascending order or descending order and totaled.

<Step 106> A data group consisting of items as a result of aggregation and a correlation coefficient is stored in the file 701 in the table format of FIG. Correlation table 701 shown in FIG.
Is based on the related level item file of FIG.
It is automatically created by referring (searching) and checking the correlation table 402 in (b).

This file 701 has a "result" field in the first field and a "cause n" in the second field and below.
(N = 1, 2, ...) ”field.

As shown in FIG. 7, for example, the item No. in the "Result" field of the first field. The item having a correlation with item 1 is the item number in the "cause 1" field of the second field. 1 or later. Furthermore, cause 2 of the third field
The field is a further causal item for each item in the cause 1 field of the second field. For example, the item N in the "cause 1" field of the second field
o. The item having a correlation with item 1 is the item number in the "cause 2" field of the third field. It is 1-2. Similarly, the item No. in the “Cause 1” field will be described below. The item having a correlation with item 2 is the item number in the "cause 2" field. 1-3
Is.

The correlation coefficient in FIG. 7 is a correlation coefficient with the item to be correlated.

<Step 107> The above-mentioned items and correlation coefficients are stored in the storage device 302 in the form of a file 701 and, if necessary, output to the display 305 in the same format.

According to this example, as shown in FIG. 7, since the related levels are input in the table format, it is possible to easily correct and reuse the levels.

Next, a second concrete example of the above embodiment will be described with reference to FIGS. 4 to 8 as in the first concrete example. Before that, referring to FIG. 3, according to the processing flow of FIG.
The operation of this embodiment will be described.

Steps 201 to 206 of FIG.
The steps up to are the same as steps 101 to 106 in FIG. Next, the items are tabulated in the correlation diagram (step 207). Step 208 is step 1 of FIG.
The same as 07.

Here, a second specific example of the data processing method based on the above embodiment will be described. Here, each processing step of FIG. 2 will be described in detail with reference to FIGS. 4 to 8. 4 to 7 are as described above. FIG. 8 is a diagram showing an example of a file format for storing data after totalization.

<Step 201> to <Step 206>
This is the same as the first specific example of the above embodiment.

<Step 207> A data group consisting of items and correlation coefficients is stored in the file 801 in the format shown in FIG.

This file is an aggregate of items for each field in the correlation table of FIG. 7 in a tree format. Each tree consists of a data group consisting of items and correlation coefficients. Each data group is linearly combined based on the correlation table.

<Step 208> The same as the first specific example of the above embodiment.

In each of the above two specific examples, the item is the management index and the data such as the business performance evaluation item or the business performance item and the data, but the attribute items (the age, the savings balance) are described. ) And data (numerical data such as monetary amount and interest rate, character code data), time items (year, month, day) and arbitrary items such as data, and data.

In the above embodiment, an example of a financial institution is shown, but the present invention is not limited to this, and in general, when performing performance evaluation in various companies such as trading companies and service industries, the performance changes It can also be used to elucidate the cause (reason) of.

[0047]

As described in detail above, according to the present invention, a correlation table storing a correlation coefficient of a single correlation between items is searched based on a plurality of items arbitrarily designated by the user. By doing so, we have created a correlation table or a correlation diagram with multiple correlations between these multiple items, so for items that the user freely selects, some items that have a causal relationship and further those items Items that are causally related to each other can be reliably found. Further, there is an effect that a plurality of sets of single correlations can be freely and efficiently combined from the correlation table according to the instruction of the operator, and all the items related to each other can be surely found.

Further, by correlating a plurality of such designated items into related levels and searching the correlation table based on the table or file in which the levels are classified, the correlation related table (figure) can be automatically created. Also, there is an effect that this kind of search and aggregation processing can be performed efficiently in a strictly short time for each level without troublesome manual work, and the accuracy of aggregation can be improved.

Further, since the user can freely set and input the range of the correlation coefficient used for the totalization, for example, the minimum value, the effect that the totalization can be freely obtained is obtained.

[Brief description of drawings]

FIG. 1 is a diagram showing a processing flow according to an embodiment of the present invention.

FIG. 2 is a diagram showing a processing flow according to a modified example of the embodiment of the present invention.

FIG. 3 is a configuration diagram of a processing device for carrying out the present invention.

FIG. 4 is a diagram showing an example of a file format for storing data before editing and a correlation table respectively.

FIG. 5 is a diagram showing an example of a file format for storing a related level.

FIG. 6 is a diagram showing an example of a screen for specifying processing contents.

FIG. 7 is a diagram showing an example of a table showing correlation relationships in a tabular format.

FIG. 8 is a diagram showing an example of a table showing a correlation relation in a graphic format.

[Explanation of symbols]

 301 computer 302 storage device 303, 304 input device 305 output device

Claims (4)

[Claims] 1. A means for inputting all items or items designated by a user from a purpose-specific database into a management table,
A correlation table that stores the correlation coefficient of a single correlation between each item obtained by the correlation analysis method, and a multiple correlation between the plurality of items is searched by searching the correlation table based on a plurality of items arbitrarily specified by the user. And a means for creating a correlation relation table or a correlation relation diagram which the data processing system has. 2. A plurality of items arbitrarily designated by the user are sequentially classified into levels according to mutual causal relationships between the items, and each item is classified into the level according to the correlation table or the correlation diagram. The data processing system according to claim 1, wherein the data processing system is divided and arranged correspondingly. 3. A unit for designating a correlation coefficient value of a predetermined value or more by a user, and a correlation coefficient value of the predetermined value or more among a plurality of items arbitrarily designated by the user based on the designation of the correlation coefficient value. 3. The data processing system according to claim 1, further comprising means for retrieving an item having “” and outputting the item to the correlation coefficient table or the correlation coefficient diagram. 4. A means for inputting arrangement designation information for designating whether to arrange the level-divided items in an ascending order or a descending order of levels, and the correlation coefficient table or the phase relationship according to the arrangement specifying information. 4. A data processing system according to claim 2, further comprising means for arranging items in several figures.