search menu linkedin twitter facebook cross phone check arrow-left arrow-right mail

Districon contributed to the INFORMS Conference 2018 in Baltimore

23-04-2018 - 12:00 - Nieuws

Districon joined a presentation of AIMMS during the INFORMS Conference 2018 in Baltimore.

Our colleague Sander van Lokven contributed in a presentation of Gertjan de Lange (SVP Connecting Business & Optimization @ AIMMS) at the INFORMS – 2018 Business Analytics Conference in Baltimore. Sander presented examples of 2 project cases which supported Gertjan de Lange’s presentation on:

The Intersection of OR and Data Science – Opportunities, Challenges, and Innovation

 

In the history of almost 30 years in the market, AIMMS has performed some pivotal but necessary changes, with the aim of helping AIMMS’ customers benefit more from the value of optimizing business processes. Driven by market requirements, customer feedback and innovation initiatives, AIMMS remains laser-focused on continuing to bring its customers more value and broadening awareness of the benefits of optimization. Learning from the development of the Data Science industry, AIMMS recognize a large but challenging opportunity for increasing their added value. One of the challenges is the potential conflict between the modeling paradigms and the conversations shift. As things evolve, significant implications and questions are likely to arise. The new developments in Data Science, the consumerization of IT and software as a service, force AIMMS to think differently on how they service their community of partners and customers.

Districon presented two projects which combined data science and OR to improve the quality of the decision in a practical setting. The quality of output obtained from optimization models is dependent on the quality of the data used as input (garbage in is garbage out). In many practical settings, Data Science is used to improve the quality of the input data, leading to higher quality results. The first example presented by Districon considered the fashion industry. It illustrated the use of forecasting on product group level, and it showed how continuously improving data helps adjusting this forecast. The next step in this process is to involve machine learning techniques to take underlying patterns into account which are not spotted by the currently used statistical models.

The second example (profit based business planning and scheduling @ Nampak) illustrated the use of Data Science in a decision process where the optimization output is reviewed. Output of multiple scenarios can be reviewed in order to make the best decision. An example is the comparison between a maximum profit scenario and a maximum customer satisfaction scenario. This review can be performed either automatically or manually. The review in the Nampak example was done manually, and the impact of the comparison was shown. In this case, it could lead to adjustment of constraints or the objective (e.g. Maximize the customer satisfaction, given that the profit must be within 10% of the Maximum profit output).