The reliable supply chain is the highest goal

Data scientists combine computer science and business administration with technical thinking. Prof. Alexander Löser from the Data Science Research Center at the Beuth University of Applied Sciences Berlin explains why data analysts can make predictions more accurate and supply chains more secure.

Prof. Löser: You are responsible for databases and text-based information systems at the Beuth University of Applied Sciences Berlin. You call yourself a data scientist – what exactly is that?

Data scientists are exploring the world with digital techniques and are drawing conclusions about future developments. We humans have been optimizing our processes for many centuries. Today, however, we can collect more accurate and timely raw data by digitizing it. Put succinctly, we can put the information into a machine “brain” that then makes predictions, for example. As a data scientist, I often learn from observation or simulation of the world. A concrete example is the route planning: programs calculate the way from A to B for us. If they know the current positions of all road users, they can recognize traffic jams. Then they can show us a faster way out based on knowledge of traffic situations in the past, for example that a traffic jam takes a long time there.

Supply chain: BVL study investigates trends in value chain reorganization | logistik aktuell

2017 BVL study: Changing consumer demands require even greater customer centricity along the entire supply chain.

logistik-aktuell.com

The same way, I have to think along both the business process (How do I get from A to B?) and the value chain (How is the system financed?). Feedback loops must also be taken into account, ie new and complementary data that we gain during operation. And finally, it’s also about technical and ethical aspects …

What do you mean by ethical aspects?

Well, for example, it is about collecting and evaluating personalized data – there must be proper privacy. But it’s also about machines later making decisions that directly affect people. For example, when an automated vehicle needs to change direction in a dangerous situation. Ethical considerations can, at least, help the data scientist to become aware of these decisions.

Prof. Alexander Löser

“We build a bridge between two previously separate data worlds: the public Internet and a company’s world of internal information. This creates a completely new knowledge network, similar to Wikidata / Wikipedia, but this time also adapted to the needs of German industry.”

Such digital analyzes must be ideal for logisticians, right?

Absolute. Our university has been working with Siemens on supply chain management since 2015, among others. The project is funded by the Federal Government. This involves products such as wind turbines. They depend heavily on reliable supply chains with their many individual parts. For Siemens, the reliability of the supply chain has top priority – but factory outages or delayed container ships can jeopardize production. We, therefore, help with this project called www.SmartDataWeb.de and collect and integrate data from many different channels on the current state of the supply chains. Together with our partners, we analyze and prepare them so that the current condition of the supply chain manager is displayed – and predictions about the future status are possible.

“#SmartDataWeb: This is how scientists are helping to make supply chains safer and more stable through #DataScience“

Tweet WhatsApp

How does that happen?

We use machines to collect text from many sources, such as newspapers, reports or posts on social media. This allows us to automatically link public data streams, such as news websites or social media channels, with company-internal data. With the Smart Data Web, we build a bridge between two previously separate data worlds: the public Internet and a company’s world of internal information. This creates a completely new knowledge network, similar to Wikidata / Wikipedia, but this time also adapted to the needs of German industry. The University of Leipzig is also a partner in the project and our contact to the Wikipedia world.

Collective efficiency: Mobility researchers can learn from ants | logistik aktuell

Working on efficient urban mobility, researchers take their cues from ants to develop mathematical routing algorithms.

logistik-aktuell.com

Are there actually limits to these machine methods?

Luckily yes! Because the data provide only a small copy of the world – and, therefore, always reflect the view of the image taker. This image, which I’ve seen, describes, first of all, the limits of what the algorithm can predict. At the same time, it is not the sum of the images that is important, but their variance and the variety of data sources.
For this reason, the basic data should be complete and not contradictory. This is often not the case with our projects for companies or medical institutions. Our neural networks need to be, therefore, optimized with many tricks for relatively few data – compared to the amounts of data that are available to Google. In addition, we can use methods to “generate” possibly missing data through simulation. That’s tricky and still in research.

How did you actually get to this field?

I was always happy when my Lego and other components worked together with the Märklin Railway. It’s a lot of fun to bring different things together and create value for our partners. And at the Beuth University of Applied Sciences we are in the highest competition!


Jobs of the future

In the future, a large number of data engineers will be needed to collect, index, and help analyze data by machine. Data Product Managers are in great demand. They know the platforms, can design data for complex analyzes and are, at the same time, the interface to customers.
For one year, the Beuth University of Applied Sciences Berlin has been offering a four-semester English Master program in Data Science for the Master of Science (M. Sc.). Interested parties can apply from April 15, 2018 to 22 places.

More about at http://data-science.berlin