Towards Identifying Feature Subset Selection for Mining High Dimensional Data

High dimensional data is the data which has many features. Some features might have representative characteristics that can help in reducing search space in data mining activities. Therefore it is important to identify such features. The feature subset selection can improve the performance of data mining on high dimensional data. This will help in extracting business intelligence that can help in making expert decisions. However, it is challenging task to identify feature subset that is representative of all possible characteristics. Song et al. , of late, proposed a framework that can be used to select feature subset from high dimensional data. Clustering is involved in their approach. Similarly in this paper we built a prototype system that demonstrates the feature subset selection. The application uses clustering and the results reveal that they are encouraging. The results are also compared with other algorithms like C4.5, Nave Bayes, IB1 and RIPPER.