DEVICE TO SELECT FEATURE BASED ON NEIGHBORHOOD GRANULARITY AND METHOD THEREOF

摘要

Disclosed are a device to select a feature based on neighborhood granularity and a method thereof. The method to select a feature based on neighborhood granularity includes: a step of initializing variables; a step of generating a neighborhood granule of sample data; a step of defining the neighborhood granularity of each of the sample data; a step of defining neighborhood approximations of each of the sample data by using the neighborhood granule; a step of calculating a dependency degree of a conditional attribute by using a group of the sample data and the neighborhood approximations; and a step of selecting a conditional attribute with the highest dependency degree as a candidate feature for the classification of the sample data. As such, the present invention is capable of increasing the accuracy of selecting a feature corresponding to a significant factor.;COPYRIGHT KIPO 2018