A Journey in Kaggle Competitions

Francisco Ribeiro Mansilha took part in the BUas Kaggle Club, an extracurricular set up by lecturer Alican Noyan that uses Kaggle competitions as a way into machine learning. The platform has around 530,000 active members worldwide, students and professionals alike.

The competition

Working alone, Francisco built a model on a dataset derived from deep learning training on the Predicting Critical Heat Flux data. He analysed the data, researched feature importance, and talked with other competitors to compare approaches - a large part of what makes the format useful.

He finished 43rd out of 700 teams.

Afterwards

His own verdict is that he could have picked a better-performing submission file at the close. Next he wants to go through the top performers’ submissions and refine his approach, build a predictive app, and enter more competitions - he finds them “highly motivating and rewarding” for building data science and ML engineering skills. This was his third.