Customer churn is one of the most critical challenges faced by businesses, especially in subscription-based industries like telecommunications. Understanding why customers leave and predicting churn in advance can help companies take proactive measures to retain valuable users. In this project, we explored customer churn prediction using Support Vector Machine (SVM), a powerful supervised machine learning algorithm known for its effectiveness in classification tasks.The dataset used in this analysis includes key customer attributes such as tenure, monthly charges, total charges, contract type, internet service, technical support, online security, payment method, and paperless billing. Initially, exploratory data analysis (EDA) was performed to understand the distribution of both numerical and categorical features. Histograms and boxplots revealed patterns in continuous variables like tenure and charges, while countplots helped visualize categorical distributions such as contract types and payment methods. These visualizations provided valuable insights into customer behavior and potential factors influencing churn.Since SVM is sensitive to feature scaling, the data was preprocessed using standardization techniques to ensure all features contributed equally to the model. Categorical variables, which were already encoded numerically, were carefully handled as classification inputs rather than continuous values. The dataset was then split into training and testing sets to evaluate model performance effectively.The SVM model was trained using different kernel functions, including linear and radial basis function (RBF), to identify the best decision boundary for separating churned and non-churned customers. The model achieved an accuracy of approximately 72%, indicating a reasonable ability to classify customer behavior. However, the classification report showed that the model performed better in predicting non-churn customers compared to churn customers, which is a common issue in imbalanced datasets.Further evaluation using confusion matrix and performance metrics such as precision, recall, and F1-score provided deeper insights into model strengths and weaknesses. The analysis highlighted that while the model is reliable in identifying customers who will stay, it needs improvement in detecting customers likely to leave. This suggests opportunities for enhancement through hyperparameter tuning, class balancing techniques, or trying alternative models.Overall, this project demonstrates how machine learning, particularly SVM, can be applied to real-world business problems like customer churn prediction. By combining data preprocessing, visualization, and model evaluation, we can derive meaningful insights that support strategic decision-making and improve customer retention strategies.
This Data analysis done by Ritika Deshmukh.
Google Colab Notebook :
https://colab.research.google.com/drive/1hhGLuvXDxECSF3laIXO7GUtKjL0J8ZAC?usp=sharing
For any inquiries, please write to us at info@macroedtech.com.
