Smart prediction of shipping volumes with AI-models

Predicting shipping volumes with artificial intelligence instead of intuitive prediction was the goal of the smart data experts at SDSC-BW together with the logistics and transport company LGI. The comparative analysis shows that the advanced models built by the SDSC-BW data analysts perform better than the conventional methods.

The Logistics Group International GmbH (LGI) has been using intuitive methods for planning shipping orders and volume prediction for capacity planning. The goal of this SDSC-BW potential analysis was to make more accurate shipping volume forecasts with the help of Smart Data tools. LGI provided the data analysts with data of a time span over nine years with 32 features, such as the dispatch date. The aim was to develop an intelligent early warning system for delivery. Apart from cleaning and transforming the original data, the data analysts generated additional features based on the data, such as product measures.

The team of experts trained the intelligent predictive model to predict daily, weekly and monthly forecasts of the shipping-order quantity. Various algorithms were implemented (e.g. ARIMA or the AdaBoost regression) and evaluated in order to find the best model. Not only could the experts show that the weekly and monthly data are better suited for the prediction models, but also that the complex models of SDSC-BW significantly outperformed the basic model. LGI was able to learn of the huge potential of its data for early warning system through the potential analysis and have a great interest in data analysis resources.

Data Innovation Community

Industrie 4.0

Project partners

LGI, Smart Data Solution Center Baden-W├╝rttemberg

Contact person

Mishal Benz, SDSC-BW, mishal.benz@kit.edu