Machine Learning Based Telecom-Customer Churn Prediction

Abstract

Customer churn or attrition refers to the percentage of customers who will discontinue with a company's service during a given timeframe. Churn rate is calculated by dividing the number of customers a company lost over a given period of time by the number of retained customers at the beginning of that time period. Churn prediction is a key predictor of the long term success or failure of a Business. In this research, machine learning and deep learning techniques are explored in order to predict telecom customer churn. Ubiquitous techniques like Random Forest Classifiers and SVMs are compared with relatively newer architectures like XGBoost and Deep Neural networks to classify if a customer will churn or not. The efficiency of these models are further explored by passing them through a grid search. From this experiment, it could be inferred the Random Forest model works best for this particular use case with a prediction accuracy of 90.96% on the testing data before grid searh.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 24 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.