https://www.kaggle.com/c/mercedes-benz-greener-manufacturing
This is a baseline approach to the Mercedes Challenge on Kaggle
The challenge consist on cutting the time that Mercedes-Benz manufacturing spends on the test bench by removing non relevant features
the approach used here is based on :
- Penalized regression (Lasso) to reduce the data dimension from 579 to 92 features
- 10fold Cross validation on reduced input using linear/non linear regression models scored on R square metric
- Hyperparameter Tuning through GridSearchCV on XGBoost Regressor(sklearn)