Skip to content

MathXCruz/Kaggle-Spaceship-Titanic

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Spaceship Titanic

This project shows my solution to the Spaceship Titanic Kaggle competition.

I'm updating this repo with every major update I make, to record how I got to each score.

  • Attempt 1 [Score = 0.80243]: the focus was 100% on the EDA and Feature Engineering. I used a GradientBoostClassifier with no hyperparameter tuning.

  • Attempt 2 [Score = 0.80336]: Added a few more ML models to compare the results and also hyperparamter-tuning with cross validation for the model that best performed.

  • Attempt 3 [Score = 0.80406]: Added LGBM and model stacking, with some new features and deeper feature engineering. The model started to overfit a lot, with ~.9488 being my best local validation score, so I started removing some variables to try to addres this problem.

  • Attempt 4 [Score = 0.80967]: Added pandas profiling for variable exploration, played with features used in the model, fixed stackingclassifier validation.

Next steps: possibly adding a NN.

About

My Kaggle Spaceship Titanic competition solution.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published