Prediction of the Least Principal Stresses Using Drilling Data: A Machine Learning Application