Project based on the application of distinct classification algorithms in order to determine the cause of wildfires.
- LDA, with a previous PCA to train with the principal components
- Tree Based Models:
- Random Forest
- Boosting
- XGBoosting
- KNN
Boosting has been the best performing model. The following images show the confusion matrix of the test set prediction, and the importance of the model's features for the best model.