This is a machine learning project that aims to predict the churn rate of a telecommunications company's customers.
The dataset used in this project is the Telco Customer Churn dataset from Kaggle, which contains information about customers who have either churned or stayed with the company. The dataset contains 21 features including customer demographics, account information, and services subscribed to.
The project uses a supervised learning approach and various classification algorithms were tested including logistic regression, decision trees, random forests, and support vector machines. Hyperparameter tuning was performed on the best performing model using grid search.
The model was evaluated using accuracy, precision, recall, F1 score, and ROC AUC. Since the dataset is imbalanced, with only about 26% of the customers churning, the models were also evaluated using the area under the precision-recall curve (PR AUC) to give more weight to the minority class.
The best performing model was a random forest classifier with an accuracy of 80%, F1 score of 0.58, and PR AUC of 0.57.
-telco_churn.ipynb: Jupyter notebook containing the code for the project
-telco_data.csv: dataset used in the project
-README.md: this readme file
The project requires the following Python libraries: pandas, numpy, matplotlib, seaborn, scikit-learn, xgboost.
The dataset used in this project was obtained from Kaggle: https://www.kaggle.com/blastchar/telco-customer-churn
Géron, A. (2019). Hands-on machine learning with Scikit-Learn, Keras, and TensorFlow: Concepts, tools, and techniques to build intelligent systems. O'Reilly Media, Inc.
Brownlee, J. (2020). How to Develop a Baseline Model for Telecommunications Churn. Machine Learning Mastery. https://machinelearningmastery.com/develop-baseline-model-performance-for-telecom-churn/