Freelance Projects

  • Dashboard implementation in Power BI.
  • Data collection through API’s and web-scraped data.
  • Report automation.

Delay prediction in Belgium railway system

Technologies & Libraries: Python, Sckit-learn, matplotlib, seaborn

  • Obtained a RMSE of 7 seconds using XGboost regressor, feature engineering, and hyperparameter tuning.
  • Developed the whole project from scratch: No background information or problem statement.
  • Achieved DataCamp's Professional Data Scientist Certification after submitting a technical report and presenting findings to DataCamp's DS experts.

End-to-End Churn Prediction

Technologies & Libraries: Python, Sckit-learn, matplotlib, seaborn, Streamlit, Heroku

  • Reached a ROC-AUC score of 99.98% in Cross-Validation and 94.01% in test-set applying Random Forest classifier.
  • Performed exploratory data analysis and data consolidation for 5 datasets with 5+ million records.
  • Deployed a machine learning web-app: https://churn-web.herokuapp.com//

End-to-End Handwritten prediction

Technologies & Libraries: Python, Sckit-learn, TensorFlow, Keras, matplotlib, seaborn, HTML, CSS, JS, Flask

  • Obtained a 98.46% accuracy to predict handwritten digits (MNIST dataset) through Artificial Neural Networks.
  • Used TensorFlow and Keras to train the model.
  • Deployed an ML App predict digits written by people