Home Credit Default Risk Kaggle submission

My submission for the Home Credit Default Risk Kaggle competition.

The objective of this competition is to predict how capable each applicant is of repaying a loan. I started with a simple exploratory data analysis investigating the distributions of features between classes, looking for correlations. Later, I experimented with manual and automated feature engineering, comparing the performance of various machine learning models.

Technology

• Python - The programming language for of this project
• Pandas and Numpy - Data wrangling,
• Various Scikit-learn tools and pipelines - Data preprocessing,
• Imbalanced-learn - Class balancing,
• Seaborn and Matplotlib - Data visualization,
• Scikit-learn, LightGBM - Machine learning model training and evaluation