Customer Churn Prediction with Machine Learning

Demo

Description:

Skills: Python, OpenAI, Machine Learning, Model Inference, Feature Engineering, Data Visualization

Part 1:

• Downloading the dataset
• Understanding the dataset
• Pre-processing the dataset
• Training the ML models
• Evaluating the ML models

Part 2:

• Building the web app

  • Creating the UI and charts
  • Making predictions with the ML models
  • Generating personalized emails with OpenAI or LLama 3.1 via Groq

  Troubleshooting:

  • Update on storing serialized model and memory snapshots using Pickle Use save_model instead of pickle.dump to save the model https://xgboost.readthedocs.io/en/stable/tutorials/saving_model.html

Developer's Notes

Add the git repo:

git init
git add *
git commit -m 'Initial commit'
git branch -M main
git remote add orgin https://github.com/itancio/<repot>.git
git push -u origin main

Create Python virtual environment

python3 -m venv venv
source venv/bin/activate

Running streamlit

cd streamlit
streamlit run main.py

Troubleshooting:

• ERROR: The error was related to XGBoost failing to load due to the missing OpenMP runtime library (libomp.dylib) on macOS

SOLUTION: Update libomp

brew install libomp

• ERROR: Handling large files

git lfs install
git lfs track "*.csv"
git add .gitattributes

git add notebook/fraud/fraudTest.csv
git add notebook/fraud/fraudTrain.csv
git commit -m "Add files"
git push origin HEAD:fraud