JPH07141187A - Data analysis supporting system and its method - Google Patents

Abstract

PURPOSE:To easily support a procedure guide in the case of reanalyzing past executed processes in the reverse order by registering plural selection elements in their priority order when mutually exclusive selection elements are simultaneously estimated. CONSTITUTION:When plural mutually exclusive factors are estimated at the time of selecting listed selection elements by an analytical result input means 306, selection elements are successively selected from the one with the highest possibility in the priority order. Selected selection elements are stored in a trace data file 313. The selection element with the highest priority out of stored selection element candidates which are not erased yet is selected and an analytical procedure introducing means 303 infers the succeeding process based upon the selected element. At the time of ending a series of analytical processes, a cause inferring means 308 infers a cause and a counterplan. In the case of executing reanalysis after inferring the cause, a procedure guide for verifying the past executed processes in the reverse order can also be supported and even an unskilled person can quickly and acurately execute data analysis.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention extracts necessary data from one or more information databases and analyzes it to confirm the existence of a problem and at the same time trace the cause of a problem and to devise a countermeasure. The present invention relates to a data analysis support system and method in a business that does business.

[0002]

2. Description of the Related Art Conventionally, some data analysis support systems for supporting data analysis by a computer for solving various problems have been developed, and a knowledge base-based expert system guides a problem solving procedure interactively. And, it is mainly classified into the system that can easily retrieve the data necessary for the work in the data-based system. The present invention improves on the former system. As an example of a conventional expert system of this kind, Japanese Patent Laid-Open No. 63-63
As disclosed in Japanese Unexamined Patent Publication (Kokai) No. -204118, there is a method in which a failure diagnosis is advanced based on a diagnosis rule and a cause is inferred from an obtained result based on an inference rule.

However, recently, a device having a function of appropriately presenting necessary data, text or image data in guiding a problem solving procedure has been developed.

FIG. 3 is a flowchart showing the operation of the conventional data analysis support system. FIG. 4 is a diagram showing an example of a decision tree in which the factors analyzed by the data analysis support system are developed in a tree structure, and is a diagram showing a decision tree assuming a data analysis operation of a quality situation in a production factory.

In FIG. 3, in step 201, for example, "process / equipment factor related to quality" is selected from the candidate group as the analysis theme, and the target period, target model, etc. (for example, model B shown in FIG. 4) are input as initial information. . When the input is complete, the system selects the first analysis process as the next process in step 202. In step 203, the status of the next process is checked from the master information file, and if the next process is an automatic analysis process, in step 204, data collection / processing / analysis is automatically performed as the next process (for example, model B), and in step 205. Automatically list the factors that could cause a problem. For example, steps 2 and 3 of manufacturing steps 1 to 3 of model B
Steps 2 and 3 are listed up if the yield at 1 is particularly low.

If the next process is a manual process, the system collects, processes, and outputs the necessary data based on the definition of the master information file in step 206, and the operator analyzes it based on the data, and selects the factor in step 207. List from the technique. Next, in step 208, a selection technique that matches the analysis result is selected from the listed selection techniques for both automatic determination and operator determination. In this case, if a plurality of mutually exclusive candidates are estimated, any one of them must be selected, and the other candidates of the selected technique listed in the system processing are rejected.

Based on this selection technique, the system infers and determines the next analysis process, and step 2 if analysis continues.
Return to 02. In step 210, it is judged whether or not the series of analysis processes have been completed, and if there is still analysis, the process returns to step 202. When the series of analysis processes is completed, the cause / countermeasure is inferred in step 211, and if the operator is satisfied with this result in step 212, step 2
Data is output at 14 and 215, and the process ends. If the operator is not satisfied with this result, the process is traced back to the reanalysis in step 213. However, the reanalysis is not supported by this conventional system, and henceforth, it has been rejected until now. In order to verify the selection technique of the process, reanalysis will be repeated so as to minimize duplication at the operator's discretion.

[0008]

In the above-described conventional data analysis support method, since it is possible to register only one selection technique in step 208 when mutually exclusive selection techniques are estimated in steps 205 and 207, There was a drawback that the solution by another series of selection techniques was rejected, and it was difficult to assist the procedure guidance when retrospectively reanalyzing the process performed up to that point.

[0009]

The present invention is a data analysis support for extracting necessary data from one or more external databases and analyzing it to confirm the existence of a problem and at the same time trace the cause when a problem exists. In the system, a master information file that stores master information, an analysis data storage file that stores data collected from an external database, a trace data file that stores initial information and analysis results of each analysis process, and analysis factors. A knowledge base for storing analysis procedure guidance knowledge and cause inference knowledge for inferring an analysis process in accordance with a knowledge expression format in which another selection technique is developed, and initial information input means for allowing an operator to input the initial information including an analysis theme. Based on the analysis procedure guidance knowledge, the initial information and the analysis result of each analysis process should Analytical procedure guiding means for inferring a process, analytical result inputting means for prioritizing and selecting a selection technique from results analyzed in the analytical process, and each analytical process obtained from the initial information and the analytical result inputting means. The analysis process trace means for storing the analysis result in the trace data file and the cause of the problem from the analysis result for each analysis process stored in the trace data file by the analysis process trace means based on the cause inference knowledge. It is configured to include cause inference means for inferring.

In the data analysis support method of the present invention, necessary data is extracted from one or more external databases and analyzed to confirm whether there is a problem and at the same time trace the cause of the problem when the problem exists. Select the first analysis process from the initial conditions including the analysis theme input by the operator according to the knowledge expression format in which the selection technique is developed,
The selected analysis technique is executed by executing the selected analysis process, the selected technique is stored in the trace data file, and the selected technique having the highest priority which is stored in the trace data file and has not been erased yet is selected. If necessary, the analysis process having the highest priority selection technique selected according to the knowledge format is selected as the next analysis process, and the highest priority selection technique is deleted from the trace data file. If selection and execution are repeated and all necessary analysis processes have been executed, after further inferring the cause of the factors indicated by the selection technique selected in the analysis process, if necessary, the trace data file is again created according to the knowledge representation format. An analysis process that has one of the stored stored selections that has not yet been written to the next analysis process. And performing the selected re analyzed as.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings.

1 and 2 are a block diagram showing the configuration of an embodiment of the present invention and a flow chart showing the operation thereof, respectively.

In FIG. 1, the execution control means 301 is an initial information input means 302, an analysis procedure guidance means 303, an information retrieval means 304, an information output means 305, an analysis result input means 306, an analysis process trace means 307, a cause inference means. The respective blocks of 308, the data analysis knowledge storage means 309, and the database management means 310 are controlled. The database management means 310 includes a master information file 311, an analysis data storage file 312, and a trace data file 3.
13, manages the knowledge base 314, and performs input / output processing of a database composed of these files in response to a request from another block.

The master information file 311 stores various master information necessary for analysis. Analysis data storage file 3
12 stores the data collected from the external database in each analysis process. The trace data file 313 stores the initial information input by the initial information input means 302 and the analysis result in each analysis process obtained from the analysis result input means 306. The knowledge base 314 stores the knowledge necessary for the guidance of the analysis procedure and the reasoning of the cause, and the data regarding the decision tree shown in FIG. 4 is also stored in the knowledge base 314. However,
These pieces of knowledge are expressed in a knowledge expression form that combines various forms such as a decision tree, a matrix, and if then rules as necessary.

The initial information input means 302 asks the operator to input the analysis theme selection and initial information, and the trace data file 313 via the database management means 310.
To store. The analysis procedure guiding means 303 infers the next analysis process from the analysis results so far stored in the trace data file 313 and the analysis procedure guiding knowledge stored in the knowledge base 314. The information retrieval means 304 collects data necessary for the analysis process from an external database based on the definition of the master information file 311, processes it into a required format, and stores it in the analysis data storage file 312 via the database management means 310. . The information output means 305 reads the data necessary for the analysis process from the analysis data storage file 312, processes it into a graph, table, etc. defined in the master information file 311 in advance and outputs it.

The analysis result input means 306 selectively inputs the result of the analysis process based on the definition of the master information file 311, and stores it in the trace data file 313. At this time, when a plurality of mutually exclusive factors are estimated, it is possible to specify from the most probable selection technique in the descending order. The analysis process trace unit 307 stores the initial information and the analysis result of each analysis process obtained from the analysis result input unit 306 in the trace data file 313. The cause inference means 308 infers the problem cause from the analysis result of each analysis process acquired from the analysis process trace means 307 based on the cause inference knowledge. The data analysis knowledge accumulating unit 309 adds / corrects the knowledge necessary for the analysis unit forward guidance and the cause inference, and stores the knowledge in the knowledge base 314.

Next, the flow of this embodiment will be described with reference to FIG. 2 by taking as an example the data analysis work of the quality situation in the production factory according to the decision tree of FIG.

In step 101, initial information input means 302
Select “process / equipment factor related to quality” from the candidate group as the analysis theme using, and input the target period, target model, etc. (for example, model B) as initial information. When the input is completed, in step 102, the analysis procedure guiding means 303 selects the first analysis process as the next process. Step 10
In step 3, the status of the next process is checked from the master information file 311. If the next process is an automatic analysis process, in step 104 the data retrieval / processing / analysis of the next process is automatically performed using the information retrieval means 304 and step 105. Then, the information output means 305 automatically lists factors (problem processes).

In step 106, the analysis result input means 306 selects the selection technique listed in step 105. At this time, if a plurality of mutually exclusive factors are estimated, the selection technique having a higher possibility is given priority. To select. The selected technique is stored in the trace data file 313. If the next process is a manual process, in step 107 the necessary data is collected and processed by the information retrieval means 304 based on the definition of the master information file 311, and the information output means 305 edits it into the yield graph for each process, for example. To display. Based on this, the operator analyzes and in step 108, factors (problem process) are listed from the selection technique as the analysis result. The operator selects the selection technique specified in step 109, but in this case as well, when a plurality of mutually exclusive factors are presumed as in the case of the automatic analysis, priority is given to the selection technique having a high possibility. The selected selection technique is stored in the trace data file 313.
In step 110, the selection candidate stored in the trace data file 313 and having the highest priority, which has not yet been deleted, is selected, and based on this, the analysis procedure guiding means 303 infers the next process in step 111. .

In step 112, it is judged whether or not a series of analysis processes have been completed, and if analysis is still left, the process returns to step 102, and an analysis process for analyzing the selected technique based on the analysis procedure guidance knowledge is inferred. Then, the selected selection technique candidate is erased from the trace data file 313. When the series of analysis processes is completed, the cause inference means 308 infers the cause / countermeasure in step 113, and when the operator is satisfied with this result in step 114, outputs data in steps 117 and 118 and ends.
If the operator is not satisfied with this result, step 11
In step 5, it is determined whether or not reanalysis is to be performed. If not, the data is output in step 118 and the process ends. If reanalysis is to be performed, in step 116, an appropriate one as the next process is deduced from the processes having the unselected selection technique candidates, and the process returns to step 102.

[0021]

As described above, the data analysis support system of the present invention can register a plurality of selection techniques in priority order when mutually exclusive selection techniques are simultaneously estimated in each process for problem solving. Therefore, when re-analyzing after inferring the cause, it is possible to support the procedure guidance when retroactively verifying the process performed up to that time, and the effect that even an unskilled person can perform data analysis quickly and accurately is there.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

FIG. 2 is a flowchart showing the operation of the embodiment shown in FIG.

FIG. 3 is a flowchart of a conventional data analysis support method.

FIG. 4 is a tree structure diagram showing a relationship between factors analyzed by the data analysis support method.

[Explanation of symbols]

 301 Execution Control Means 302 Initial Information Input Means 303 Analysis Means Guidance Means 304 Information Retrieval Means 305 Information Output Means 306 Analysis Result Input Means 307 Analysis Process Trace Means 308 Cause Inference Means 309 Data Analysis Knowledge Accumulation Means 310 Database Management Means 311 Master Information File 312 Analysis data storage file 313 Trace data file 314 Knowledge base

Claims (4)

[Claims] 1. Master data is stored in a data analysis support system that extracts necessary data from one or more external databases and analyzes it to confirm whether there is a problem and at the same time traces the cause when a problem exists. Master information file, analysis data storage file that stores the data collected from the external database, trace data file that stores the initial information and the analysis result of each analysis process, and the knowledge that the selection technique that is the factor of analysis has been developed. A knowledge base for storing analysis procedure guidance knowledge and cause reasoning knowledge for inferring an analysis process according to an expression format, initial information input means for allowing an operator to input the initial information including an analysis theme, and the analysis procedure guidance knowledge. Based on the initial information and the analysis result of each analysis process, the analysis method that infers the analysis process to be advanced next. Forward guidance means, analysis result input means for selecting a selection technique by prioritizing it from the results analyzed in the analysis process, and analysis results in each analysis process obtained from the initial information and the analysis result input means And a cause inference means for inferring a problem cause from the analysis result of each analysis process stored in the trace data file by the analysis process trace means based on the cause inference knowledge. A data analysis support system characterized by including and. 2. The data analysis support system according to claim 1, further comprising data analysis knowledge storage means for additionally correcting the analysis procedure guidance knowledge and the cause inference knowledge and storing the knowledge in a knowledge base. 3. A data analysis support method for extracting necessary data from one or more external databases and analyzing the data to confirm whether there is a problem and at the same time tracing the cause of the problem when the problem exists. The first analysis process is selected from the initial conditions including the analysis theme input by the operator according to the knowledge expression format in which the selection technique is developed, the selected analysis process is executed, priority is assigned, and the selected selection technique is traced. Analysis with the highest priority selection tricks stored in the data file and stored in the trace data file and not yet cleared and, if necessary, selected according to said knowledge form Select the process as the next analysis process and delete the highest priority selection technique from the trace data file and analyze it. If the selection and execution of processes are repeated and all the necessary analysis processes have been executed, after further inferring the cause of the factor indicated by the selection technique selected in the analysis process, if necessary, the trace data will be again according to the knowledge expression format. A data analysis support method characterized in that an analysis process having one of the selection techniques stored in a file and not yet erased is selected as a next analysis process and re-analyzed. 4. The analysis process is classified into an automatic analysis process and a manual analysis process, and the automatic analysis process is automatically executed until selection of a selection technique, but in the manual analysis process, an operator selects the selection technique. Item 3
Described data analysis support method.