Exploratory Analysis of Customer Churn: Utilizing Visualization Tools Available for Python

Abstract

The role of Exploratory Data Analysis (EDA) proved to be important in understanding customer churn in telecommunications by examining a dataset of over 7000 customers, utilizing tools from the Python ecosystem, particularly Jupyter Lab, along with libraries such as pandas, NumPy, Seaborn, and Plotly. Based on John Tukey's foundational concepts of EDA, this paper emphasizes the importance of visualizing data to uncover patterns and relationships that influence customer behavior. By systematically analyzing various demographic and service-related attributes, we identified significant trends, including the disproportionate impact of contract types, customer demographics or marital status, and price sensitivity on churn rates. Obtained results highlight that customers with month-to-month contracts, and those without partners or children, exhibit higher churn tendencies. The generated interactive visualizations provide not only intuitive insights but also assist in a deeper exploration of hidden anomalies and trends, setting a robust groundwork for predictive modeling. Ultimately, this analysis underlines the necessity of EDA in formulating effective customer retention strategies, offering ways for future research employing advanced analytical techniques like machine learning and cohort analysis to predict and mitigate churn.