First place - NDD Tech ML Competition 2022

After a two year hiatus from kaggle-like competitions while I focus on my PhD, happy to have won first place on NDD Tech’s Machine Learning Competition!

I developed the highest accuracy solution for classifying NF-e’s (“Nota Fisacal Eletrônica” - Brazil’s eletronic invoice/payment system) as suspicious or not. It was interesting to develop a solution that worked directly with the raw data instead of a already “clean” dataset - The input was the XML files which businesses report directly to the fiscal authorities and it’s parsing and feature engineering challenges were directly relevant to a good solution.