In this competition, we were asked to predict the tumor size based on certain features present in the dataset. The feature engineering took me to a rank of 15 alone and finally with an ensemble of two CatBoost models a leaderboard rank of 8 was achieved.
With extensive EDA I found repeated entries in train and test which were directly replaced in post processing to improve scores. Furthermore, dropping rows from train dataset with 0 as target values improved score.
