A class of statistical procedures that have become very important in recent years are predictive methods, often called machine-learning algorithms. The basic goal of this methods is to to be able to make predictions from a given dataset with the lowest possible error. In doing so, they typically use relatively complicated, often non-parametric methods that typically don’t allow to calculate inferential products such as the MLE or p-values.