Training to the test set is a type of overfitting where a model is prepared that intentionally achieves good performance on a given test set at the expense of increased generalization error. It is a type of overfitting that is common in machine learning competitions where a complete training dataset is provided and where only … Continue reading How to Train to the Test Set in Machine Learning