The aim of this project is to perform a supervised classification of customers as to their probability of churning. To make our predictions we use historical transactional data from the past two years.

Most of the preprocessing steps can be accessed from the classes defined in src. The user should run the preprocessing steps and then the extract_labels.ipynb notebook. All the modelling is done in the model.ipynb notebook.

Once the model has been trained and that all files have been saved in the data folder, the user can run a minimal streamlit application by running from the root directory:

streamlit run src/app.py

The user can install all the required packages by running from the root directory:

pip install -r requirements

The python version used for this project is 3.8.12