A Comparative Study Based on Rough Set and Classification Via Clustering Approaches to Handle Incomplete Data to Predict Learning Styles

Handling of missing attribute values are a big challenge for data analysis. For handling this type of problems, there are some well known approaches, including Rough Set Theory (RST) and classification via clustering. In the work reported here, RSES (Rough Set Exploration System) one of the tools based on RST approach, and WEKA (Waikato Environment for Knowledge Analysis), a data mining tool—based on classification via clustering—are used for predicting learning styles from given data, which possibly has missing values. The results of the experiments using the tools show that the problem of missing attribute values is better handled by RST approach as compared to the classification via clustering approach. Further, in respect of missing values, RSES yields better decision rules, if the missing values are simply ignored than the rules obtained by assigning some values in place of missing attribute values.