Tuning the number of estimators, learning rate, and max depth for the base estimator tree has improved the model. Now, I need to investigate whether or not that the model is over-fitting the training data. New submission to the Kaggle competition is in the works.

Accuracy results for n_estimators = 5

Accuracy results for n_estimators = 10

Accuracy results for n_estimators = 15

Accuracy results for n_estimators = 20

Accuracy results for n_estimators = 25

Accuracy results for n_estimators = 30