Prediction of Cold Chain Transport Conditions Using Data Mining

Abstract

Ensuring the delivery of temperature-controlled products in transportation is an increasing challenge, especially in countries with continental extension and tropical climate such as Brazil. Products with this type of specificity generally have a higher added value and involve specialized equipment and labor. Thus, route mapping is necessary for the logistics of the cold chain. The study aimed to predict the transport conditions in the cold chain. The data set analyzed includes the temperature of the loads and the route information (Southeast to Northeast and South of Brazil). The classification of temperature excursions considered data below 15 °C or above 30 °C. The Naïve Bayes and Multilayer Perceptron algorithms are used to predict the optimal temperature excursion model. The Multilayer Perceptron algorithm proved to be the most suitable for a thermal route mapping model. With this identified standard, logistics decision making can be improved to reducer o waste and ensure product integrity with less recourse.