During this project I will create a model that predicts whether or not a customer will churn. Churn occurs when customers discontinue or cancel their subscription to a service, and it is a major issue for telecom companies such as ours. Loss of customers can result in significant revenue loss and can even harm customer loyalty. However, with our model we will address this issue by predicting which customers are likely to churn and taking proactive measures to keep them.
- We have to install Anaconda and python 3.10.9 in order to ensure that basics packages are installed like numpy, pandas, etc.
Please download the two datasets in my google drive and unzip the file using 7-Zip File Manager to retreive the CSV files.
- train_dataset_churn.csv file : Used to train our model
- test_dataset_churn.csv file : New data to predict using the built model
At the end of this project, we succeeded to put in place a model able to predict to churn status with an acceptable level of evaluation metrics focused on "precision" and "recall" (85% and more). We also written a blog post for a technical audience in order to present our work :
https://medium.com/@kowapatrick/customers-churn-predicting-in-telecom-company-fc1fd8ef2771
My express gratitude and recognition to Khanyi Noganta (https://www.linkedin.com/in/khanyisa-noganta) for the support, advice and guidance during this project.