Machine Learning Methods for Predictive Customer Churn Analysis in the Telecom Industry

To examine customer churn in the telecom sector, we used four classification algorithms: Decision-Tree(DT), Random-Forest(RF), Logistic-Regression(LR), and a Hybrid-Algorithm(HA). The dataset contained information about telecom users, including demographics, use trends, and customer service data. By resolving missing values and outliers during the preprocessing of the data, we were able to comprehend the variable distribution and connections. The four classification models were then applied to the dataset by dividing it into training and testing sets. The models' accuracy outcomes were as follows: The accuracy rates for Logistic-Regression (LR), Decision-Tree (DT), Random-Forest (RF), and the Hybrid Algorithm(HA), which combines DT and RF capabilities, were 85%, 91%, 94%, and 95%, respectively. Utilising measures like Accuracy, Precision, Recall and recall rate, we evaluated the models' performance. Our research showed that the Random-Forest (RF) model performed best and had the highest degree of accuracy. The hybrid approach greatly increased classification accuracy, underscoring its usefulness. The Hybrid Algorithm(HA) may be used by telecom businesses to estimate customer attrition and aggressively retain clients. These findings highlight how crucial it is to use the right algorithms for accurate customer attrition prediction, with the Hybrid Algorithm (HA) offering a practical way to raise classification accuracy.