A decision-making framework with machine learning for transport outsourcing based on cost prediction: an application in a multinational automotive company

Organizing decision-making processes in companies so that they are well-structured and consistent is very important in the constant search for competitiveness and sustainability in business. A recurring and relevant problem refers to the selection of suppliers for outsourced processes, as is the case of outsourcing transportation. In this context, this manuscript presents a model to help managers select freight companies, based on the assessment of logistics costs, applying Machine Learning techniques. The model is integrated with a Decision Support System and was applied to a real case of a multinational automotive company in Brazil, comparing the results with what occurred in practice. The results showed that the automotive company could have saved approximately 7% of its logistics costs by shipping its products annually, with a confidence level of 95%. The proposed framework showed advantages for the company, such as the possibility of quickly simulating possible scenarios and mitigating the logistics costs involved.