Evaluation of machine learning methodologies to predict stop delivery times from GPS data

•Regression models (RM) and classification models are be used to predict the stop delivery time (SDT).•Based on MAPE, hazard duration models (HDM) are recommended over RM for SDT predictions.•K-nearest neighbor (KNN) provides the best results when predicting if SDT exceeds a time threshold.•A two level model (classification -KNN- and regression/hazard duration) improves the prediction the stop delivery time (SDT).