Project Details

Description

The project revolves around a collaboration between NetHire and DTU Compute under the Smart Innovation initiative. The company is hiring out products (equipment and machines) to the construction industry from various locations that are scattered across the country.
Our primary focus has been to predict the demand for various types of products and to calculate the associated inventory. Since the demand fluctuates over time and the products return to inventory after a certain period, the inventory calculations are based on a time-dependent Markov chain formulation of a queueing system with customer rejection. Demand predictions are based on a heuristic that creates a weighted average of several Poisson regressions.
Our results have been applied as Java code and subsequently implemented in the company products.

Article about the project (in Danish): https://www.mestertidende.dk/article/view/653396/indkobsbeslutninger_kunstig_intelligens_afloser_mavefornemmelse
StatusFinished
Effective start/end date01/01/201926/03/2019