# Top 1 Solution

Hola/Hello everybody,
First of all, thank you for hosting the competition, was super fun to compete.
There is no magic int terms of what I have done to win this, I have tried many algorithms and the one that worked the best was XGBoost.
The solution: 5 folds XGBoost with tuned up parameters -> take the ensemble mean of all 5 predictions.
I saw a lot of people dropping the features that were missing numbers, that is a wrong way to deal with it in this competition. XGBoost actually allows you to take in the NaN.
Best regards and congratulations to all the participants !
As long as you learned something new, you are a winner.