Proposed Combined Technique of Statistical Filter and Machine Learning for Exploratory Data Analytics and Features Selecting of Telecommunication Customer Churn

Abstract

This study can determine a customer churn based on his historical data and behavior. It indicates that an efficient churn prediction model should employ a significant volume of historical data to identify churners. However, existing models have several limitations that make it difficult to do churn prediction reasonably and accurately. To solve this issue this study proposed new combined technique of statistical filter and machine learning preprocessing is used. Furthermore, statistical methods are utilized to generate models, resulting in poor prediction performance. Also, benchmark datasets are not employed in the literature for model evaluation, resulting in a poor representation of the actual visual representation of data. Without benchmark datasets, it is impossible to compare different models fairly. An intelligent model can be utilized to relieve current issues and deliver more accurate churn prediction. Keywords: Customer churn, Data Acquisition, Exploratory Data Analytics, Feature selecting technique, Filter statistical technique, Machine learning.