June 14, 2018 – hosted by community partner SAP in s’Hertogenbosch

During the interviews with our community members, the Data-Driven Inventory charter unequivocally raised most interest. The event was thus focused on sharing the first research insights on this common theme.

Opening with SAP and Sarah Gelper on the Data2Move Community

After a short welcome by Jan-Theodoor Wiltschek from SAP, Sarah Gelper from TU/e caught up on Data2Move’s progress: from the brainstorming sessions during the kick-off in September 2017, via the Data Ambition Session of February, to presenting the first results today and towards defining joint projects as a next step.

The First Results! – Insightful reports from 3 research projects

Time (in)efficiencies in demand planning by Nazli Akgül (Valeant)

How can demand planners use their time more effectively to obtain better forecasts and reduce backorder value? Using the Lean Six Sigma methodology, Nazli Akgül found that while time spent on forecasting increases accuracy, time spent on reports, meetings and data quality issues decreases accuracy. Her study also showed that time spent on backorder analysis and meetings with product managers decreases backorder value, while time spent on data quality issues increases backorder value. These results emphasize the need for integrated data quality systems.

Advised versus actual orders by Bob van Beuningen (Jumbo)

Why do store managers order different quantities than suggested by the order advise system? Bob van Beuningen found that the four most important reasons for such deviations are: second placings, changing weather, improving the store view and inventory collection. While store managers thus have good reasons to make order adjustments, these adjustments are often incorrect – especially when the actual order exceeds the adviced order. As a next step, he will investigate whether there are differences between product categories.

Starting with demand forecasting by Bas Buijse (TKT)

Which type of time series model best predicts demand for laundry? Bas Buijse conducted an extensive forecast method comparison, including external variables such as the mean temperature, humidity, sunshine duration and the GoogleTrends results of the search term ‘hotel’. His analysis shows that forecasts which take external variables into account can be very powerful and can help to better assign resources.

Bas1
Bas Buijse explaining the results of his research

Masterclass Mashup – Deep dive into one of the three key concepts in Data-Driven Inventory

Big Data to Mitigate the Bullwhip Effect by Zümbüt Atan (TU/e)

Zümbül Atan gave a masterclass on a very important phenomenon in inventory management: the bullwhip effect. There are some structural causes of the bullwhip effect, but there are also behavioral reasons caused by human decision making. She explained these causes by showing the results of a real-life experiment: the beer game. Although big data cannot entirely eliminate the bullwhip effect, integrating large and various sets of data in real time, sharing data and interacting in a collaborative manner can help to reduce the bullwhip effect.

Forecasting: Algorithm vs Human by Sarah Gelper (TU/e)

Sarah Gelper gave a masterclass on forecasting, focusing on the difference between algorithmic forecasting models and forecasting done by experts. Using her own research, other research and industry experience, she concluded that neither type of forecasting is better. Rather, a combination of algorithmic models and human expertise is virtually always the better option regarding demand forecasting.

Inventory and Transportation by Luuk Veelenturf (TU/e)

Luuk Veelenturf’s masterclass focused on transportation. The importance of online shopping over regular shopping is increasing and the role of transportation in online shopping plays crucial role. He highlighted the need for efficiency in transportation by increasing volume utilization and hit rates. In order to achieve this, the role of data science in transportation is very important. He also mentioned different kinds of data which can be used in transportation such as volume, travel & services time and delivery location. He concluded his presentation with the privacy issues in data that can be used in transportation for delivery optimization model.

Keynote Data-Driven Inventory in practice by Dori van Hulst (Pipple)

Dori van Hulst from Pipple took us on a treasure hunting trip. Just like on an actual treasure hunting trip, multiple elements are required for a successful data-science project: a clear goal, a crew, the data you want to use, knowledge and insights to gain from the data, deciding if the treasure is in line with company goals and data science techniques which are needed to find the treasure. All attendants defined these elements for their own treasure hunt, enabling them to take the first next steps.

dori2
The community filling out their own treasure hunt list

Partners on Stage – Meet three community partners and their practice

CTAC has facilitated the new online platform called ‘Movement’. The community can use this platform to share updates on research projects, discuss ideas to benefit from each other’s expertise and keep the knowledge flowing in between meet-ups. Jordi Peters pointed out the importance of working together and their willingness to collaborate.

De Persgroep is a new member of the Data2Move community. Gerda van der Poel explained the challenges De Persgroep currently faces in light of the evolving landscape of news consumption, and the way their business is changing. As exploring data can help them to set up a new logistics strategy, De Persgroep is excited to improve their distribution with the help of Data2Move.

Siel Vroman introduced the Data2Move community member Pro Alliance. She explained the benefits of Zuckerberging your supply chains, thereby introducing the theme of next event: Supply Network Collaboration.