“At an absolute minimum, a basic machine learning model should contain ten times as many data points as the total number of features.”