This project aims to predict churn using datasetss from the Telco industry
- Refine metadata stats
- Add accuracy plot in page 03
- Implement XGB modeel
- Implement Text mining
docker pull kidrissa/churnapp:latestdocker run -p 8501:8501 -d kidrissa/churnapp:latestThe project is set up in order to practice a tuto followed in an online course. Using a dataset from Telco industry the idea in the online course was to model churn by implementing an end-to-end data science project from data preprocessing, data modeling, model comparaison to putting champion model into production. The project was done using SAS viya, a low-code environment. Consequently I decide to reimplement the project using the first two C's of Cloud Native approach that is Code and Container. My aim is to build a streamlit app on which user can follow the project end-to-end by entering some hyperparams to his convenience.
One continuous integration (CI) procedure with 2 jobs mainly is designed and launched at every push and merge to the main branch:
- Linting & Testing. Flake8 grants high quality of code while Pytest collects the test from the tests folder and executes them
- if Testing goes through, a Docker Image is built and pushed onto the docker hub
The dataset used is