Cross-Sector Application of Machine Learning in Telecommunications: Enhancing Customer Retention Through Comparative Analysis of Ensemble Methods

This work investigates the Supply chain evolution over diverse industries and forecasts a telecom churn using a publicly available dataset, besides offering a thorough technique for assessing and projecting customer attrition. The study begins with a thorough literature search that examines the growth of AI in supply chain management (SCM) across several industries, including healthcare and e-commerce, in order to give a comprehensive background and contextualize the telecom sector’s accomplishments with reference to AI. The methodology includes applying different machine learning models to predict customer turnover by meticulously pre-processing the data and conducting exploratory data analysis (EDA). In order to handle missing values, a hybrid technique using K-Nearest Neighbors (KNN) Imputer for numerical features and Simple Imputer for categorical variables was used during the data preprocessing stage, which involves deleting duplicate entries and unnecessary columns. Through informative statistics and graphics, Exploratory Data Analysis (EDA) revealed important elements like rival offerings and the kind of internet service, thereby delivering insightful information regarding what causes churn. During the investigation, various machine learning models, such as Decision Tree, Random Forest, K-Neighbors, and XGBoost classifiers, were used. With an accuracy of 98.25%, Random Forest shows improved performance over the Decision Tree model, which had an accuracy of 98.02%. These models were tested using accuracy, precision, recall, F1-score, and AUC-ROC. This approach emphasizes how important predictive analytics is to comprehending the dynamics of customer turnover and further lays the groundwork for tactical interventions meant to improve customer satisfaction and retention in the telecom industry.