Delivery Time Prediction for On-Demand Delivery Services using Deep Learning

  • Hugo D. Calderon-Vilca, Gianmar H. Sanchez-Valdez, Cristina V. Caballero-Hervias, Luis A. Arce-Llantoy, René A. Calderon Vilca, Reynaldo Sucari Leon


In big cities, like the capitals of a country, logistics companies are worrying about making Just-In-Time deliveries. However, there is uncertainty in estimating travel time for in-city delivery.In this research, we propose a deep learning model to predict delivery time based on registered order data, we design a hybrid approach that contains components such as: deep feature learning, feature adaptation and classification. The model has been trained with 800 patterns and 200 tests which were taken by a logistics company. The results of the computational experiments show that the proposed model reached an accuracy percentage of 82.7%.