A project to derive banking customer ages from their demographics and products, plus the formal, product recommendation Kaggle Challenge.
capstone_writeup.ipynb - an indepth study of the Santander Kaggle dataset, with data optimisation, EDA, then customer age prediction using a variety of models, and a manually tuned product recommender.
product recommender, barebones.ipynb - a side project, based on the learnings from the main piece, to attack the Kaggle Product Recomendation challenge with as few lines as possible (100 lines of code, plus comments and spacing). Achieves a MAP@7 of 0.0259, vs. a baseline of 0.010, and a $60k winning score of 0.031