作者: Adnan Amin , Faisal Rahim , Imtiaz Ali , Changez Khan , Sajid Anwar
DOI: 10.1007/978-3-319-16486-1_22
关键词:
摘要: Predicting the behavior of customer is at great importance for a project manager. Data driven industries such as telecommunication have advantage various data mining techniques to extract meaningful information regarding customer’s future behavior. However, prediction accuracy these significantly affected if real world highly imbalanced. In this study, we investigate and compare predictive performance two well-known oversampling Synthetic Minority Oversampling Technique (SMOT) Megatrend Diffusion Function (MTDF) four different rule generation algorithms (Exhaustive, Genetic, Covering, LEM2) based on rough set classification using publicly available sets. As useful feature extraction can play vital role not only in improving performance, but also reduce computational cost complexity by eliminating unnecessary features from dataset. Minimum Redundancy Maximum Relevance (mRMR) technique has been used proposed study which selects best subset reduces space. The results clearly demonstrate both rules that will help decision makers/researcher select ultimate one.