LOGISTIC REGRESSION AND ITS USE IN BINARY CLASSIFICATION PROBLEMS SUCH AS PREDICTING WHETHER A CUSTOMER WILL CHURN OR NOT

ABSTRACT

This research explores the application of logistic regression in addressing binary classification problems, focusing on the critical issue of predicting customer churn in business settings. The study begins by defining the problem of customer churn prediction and its significance in business optimization. It discusses the challenges faced in binary classification, emphasizing data sparsity and imbalance. Throughout the research, the significance of logistic regression as a statistical tool for predicting customer churn becomes evident. The logistic function's role in modeling binary outcomes is elucidated, along with its mathematical foundations. Emphasis is placed on interpreting logistic regression coefficients and odds ratios to extract meaningful insights about the drivers of churn. The research also addresses the importance of selecting appropriate evaluation metrics and the use of crossvalidation for model assessment. Additionally, challenges related to data sparsity and imbalance are acknowledged, with proposed solutions like feature engineering and data resampling techniques. Ultimately, this study contributes to the understanding of logistic regression's role in binary classification, with a focus on predicting customer churn. It underscores the importance of statistical techniques in improving business operations and customer relationship management, emphasizing the practical implications of this research for real-world applications.