This project is a logistic regression model used to predict customer churn with an easy to use interface. It is built with strongly typed Python. This is what you’ll see the first time you launch:
This is how light mode looks:
To train the model, press “Choose XLSX File” and select the “Telco_customer_churn.xlsx” file as so:
After you do you’ll see this prompt telling you that training is complete:
In addition to the prompt, a “customer_churn_pipeline.pkl” and “customer_churn_test.pkl” file will be generated to save the model and avoid retraining every launch. This is what the interface shows after training:
It won’t let you press “Predict Churn” until all fields are filled and invalid data isn’t able to be put in any fields to prevent unwanted behavior. When you fill out the form properly and press “Predict Churn” this is what you’ll see:
Pressing “Reset Form” would set the form back to the defaults you see when you launch. This is the ROC Curve:
This is the Confusion Matrix:
Lastly, this is Feature Importance:
