Challenge
In total, several tens of millions of parts are processed each year in Mode Logistik GmbH & Co. KG's largest warehouses. Incomes fluctuate significantly due to different suppliers. Delivery behavior varies depending on the supplier, which makes planning extremely difficult.
Approach
In order to improve existing forecasting approaches, all suppliers were assessed on their previous delivery dates. Based on open orders, the corresponding delivery time windows and the expected timeliness of the suppliers, expected values of incoming goods quantities could be calculated precisely at part level. These expected values were in turn used as input for a machine learning model, which was also extended to include factors such as vacations, holidays and inventory periods in the warehouse.
Outcomes
As a result of the combination of supplier evaluation and machine learning forecast, the forecasting error in Demand forecasting Be reduced by up to 51% The results not only help the customer to better plan personnel and warehouse use, but also enable more intelligent management of suppliers in the future. For example, peaks in workload could be anticipated and smoothed out in advance by distributing orders.
*There are two independent companies Peek & Cloppenburg with headquarters in Düsseldorf and Hamburg. This reference refers to Peek & Cloppenburg KG based in Düsseldorf, whose locations can be found at www.peek-cloppenburg.de.