Sebastian1981 / Churn_Prediction_AutoML

implementation of a customer churn model using auto-ml

Churn_Prediction_AutoML

This project aims to implement a customer churn model using auto-ml including the model development, model explainability/fairness and model deployment. To get the code running, set up e.g. a conda environment using the packages in the requirement.txt file.

Step #1 - Model Development

Run the notebooks in the following order:

• "step_01_overview.ipynb"
• "step_02_exploratory_data_analysis.ipynb"
• "step_03_model_training.ipynb"
  Details on what the notebooks exactly do are provided in the notebooks itself.

Step #2 - Model Explainability and Model Fairness

Run the notebooks in the following order:

• "step_04_model_explainability.ipynb"
• "step_05_model_fairness.ipynb"
  Details on what the notebooks exactly do are provided in the notebooks itself.

Step #3 - Model Deployment

• run "step_06_web_api.ipynb" to create a fast-api python script and dockerize it
• use docker desktop to explore the app

Step #4 - Model Monitoring

• model performance and data drifts are monitored using the mlflow package
• to start the mlflow gui from the terminal, simply run $mlflow ui