MINING TELECOMMUNICATION DATA TO PREDICT CUSTOMER CHURN WITH FILTER-BASED FEATURE USING WEKA

Abstract

It is important for companies operating in the telecom field to recruit new customers while retaining their customers and avoiding losing them. There are many reasons why customers cancel their contracts, such as poor service experiences, or changing personal situations. In the past decades, the use of data mining methods for decision-making has increased in several areas, and the successful use of data mining methods has shown great advantages in many areas.This paper is aiming to build a classification model, to improve the customer churn prediction with help of feature selection techniques that focus on the most factors that affect this issue.The process of selecting influencing factors plays an important role in the data mining process, the database contains several factors that are not relevant to an effective classification process.After feature selection, this paper performs three different classifiers on the dataset to bring out the best results and compare them.The IBM Sample Data Sets was used in this technique which has 7043 user’s entries and 21 features. The performance was evaluated using Accuracy, Precision, Recall rate, and Confusion Matrix.