Machine Learning With Go

Cross validation

In addition to reserving a holdout set for validation, cross validation is a common technique to validate the generality of a model. In cross validation, or k-fold cross validation, you actually perform k random splits of your dataset into different training and test combinations. Think of these as k experiments.

Once you have performed each split, you train your model on the training data for that split, and then evaluate it on the test data for that split. This process results in an evaluation metric result for each random split of your data. You can then average these evaluation metrics to get an overall evaluation metric that is a more general representation of model performance than any one of the individual evaluation ...