Decision table data reduction method

Abstract

The invention provides a decision table data reduction method. The method comprises the following steps: step one, judging whether the last condition attribute in a decision table data set is a decision table core attribute, adding the data of the attribute in a reduction set R if the last condition attribute is the decision table core attribute, and executing the step 2; deleting the last line of condition attribute data if the last condition attribute is not the decision table core attribute, and re-executing the step one; and step two, placing the line data corresponding to the decision table core attribute in the first line, and outputting the reduction set R after determining that an end condition is met, or else returning to step one. The method provided by the invention has the advantage of simply and effectively performing reduction on the decision table data.

Description

technical field [0001] The present invention relates to the technical field of data processing, and more specifically, relates to a data reduction method of a decision table. Background technique [0002] At present, with the rapid development of data acquisition and storage technology, the problem of data redundancy is becoming more and more serious. It not only wastes storage space greatly, but also significantly reduces the performance of data-based modeling and decision-making algorithms. Rough set theory is a theory that specifically reduces data and extracts effective information from data. The core of this theory lies in data reduction. By deleting unimportant and redundant data and attributes, a streamlined new data set containing complete information is obtained, providing high-quality data for data-based analysis, modeling, decision-making, etc. source data. [0003] Traditional data reduction methods often adopt a heuristic reduction structure based on attribute...

AI Technical Summary

Problems solved by technology

[0004] However, there are some deficiencies in this traditional heuristic reduction structure, mainly in the following aspects: first, the importance is calculated too many times, step 2 will be executed many times, and most attributes will be calculated multiple times. If Step 4 adopts addition mode, then the importance needs to be calculated (2|C|-|R|+1)*|R| / 2 times, if step 4 adopts subtraction mode, then the importance needs to be calculated (|C|+|R |+1|)*(|C|-|R|) / 2 times, therefore, no matter whether the calculation formula of attribute importance is simple or not, a lot of time will be wasted; secondly, the random heuristic problem based on attribute importance , the existing attribute importance calculation methods may generate multiple attributes with the greatest importance, and the existing solutions often choose randomly in step 3, which will have a difficult impact on the result of attribute reduction and classification accuracy. predictive effects

Images