Client Defection Prediction Using Machine Learning Techniques with Data Analytics

Loss of customers remains one of the biggest issues in the industry. Understanding and predicting customers' attrition can both increase profit and enhance their retention strategy. By applying churn prediction, the industry sector can help detect potential unhappy customers at an early stage, initiate proactive measures, and retain them. The industry generates a huge amount of data due to this large customer base every day. Business analysts and decision-makers highlight that keeping the current customers is cheaper than getting new ones. Business analysts should also have knowledge of what are the causes of attrition and how are the behavior patterns found in available churn data related to those causes. This study suggests a churn prediction model that makes use of classification techniques to categorize churn clients and offer the features that lead to consumer churning in the sector. According to the results, performance metrics such as accuracy, precision, F1-Score, and recall with regard to predictive accuracy. We found that Monotonic XGBoost with Random Forest performed better than K-Nearest Neighbors, Support Vector Machines, and Decision Trees with 86% of Accuracy, 72% of Precision, 73% of recall and F1 Score of 77 %.