| Effective data mining solutions have for long been anticipated in Customer Relationship Management (CRM) to accurately predict customer behavior, but in a lot of research works we have observed sub-optimal CRM classification models due to inferior data quality inherent to CRM data set. In this thesis, I will discuss one type of CRM data with a distinctive distribution pattern which I define as the "Reduced Dimensionality". I will in term present my new classification framework, termed Partial Focus Feature Reduction, poised to resolve CRM data set with Reduced Dimensionality using a collection of efficient data preprocessing techniques characterizing a specially tailored modality grouping method to significantly improve feature relevancy as well as reducing the cardinality of the features to reduce computational cost. The resulting model yields very good performance result on a large complicated real-world CRM data set that is much better than ones from complex models developed by renowned data mining practitioners despite all data anomalies. This study opens a new way of thinking in solving CRM classification problems. |
