Abstract
The daily demand forecasts of orders in logistics businesses are critical in scheduling and planning supply chain activities in order to satisfy customer demand in a timely fashion, improve effectiveness and reduce expenses. Although some machine learning methods have been employed to predict daily demand orders of goods for the supply chain in logistics businesses, there are important concerns over the selection of the most suitable predictive technique. This paper examines the implementation of PART-based strategy to predict daily product requests in a brief period. A true database of a Brazilian logistics business gathered over 60 days evaluates the methodology. The PART classifier uses 10-fold cross-validation to forecast daily demand for orders within 6 days 10 times on this gathered dataset. The results demonstrate that the classifier suggested can predict a day-to-day order requirement with elevated precision compared to the state-of - the-art baseline classifiers.