Telecom Churn Prediction
In this project, I try to predict customers who have the potential to churn at a telecom company. Exploratory data analysis and machine learning models were built and applied to this project. First, the data must be read, either from CSV, SQL, or other forms. Then, check the data information and summary. This will help you see the entire data. I find that some columns have missing data, so we need to handle the data. If you check the total of missing data, you will find that the number is very low compared to the total data. Therefore, we can drop data without worrying about affecting the entire dataset.
After we do data exploration, we know the data and its distribution. We will move to create a machine learning model. But before that, we want to know churn's influential and essential features. We can use Mutual Information to get it. As a result, the top 5 most essential features are contracts, tenure, payment methods, online security, and internet services. Then, some machine learning algorithms are built to make churn prediction models. The best model by its classification performance is chosen for this project.