Extracting knowledge from large-scale data, is one of purpose of data mining. Since large-scale data is composed of many variables in most of cases, it is difficult to interpret the rule obtained from the data. Variable selection method have been developed to improve rule readability, however existing variable selection method would not select good variables because only each variables are considered. This paper proposed a variable selection method based on subspace comparison. Proposed method considers correlations among the variables by comparing between subspace obtained from data and basis from original data. Effectiveness of proposed method is shown through experimental result.
展开▼
