What data analytics has to do with service
Data analytics is an area that we can no longer imagine our everyday lives without. Although not always obvious, data analytics and machine learning are involved in one form or another in most of the technologies and products that surround us today. Sebastian Klemm, Head of DTO at thyssenkrupp Materials Services, talked to us about why we talk about “Data Analytics as a Service” at thyssenkrupp:
Mr. Klemm, today we are talking about “Data Analytics as a Service”. To what extent would you understand data analytics and machine learning as a service?
We understand a service to be a continuous performance that is tailored precisely to the needs of our customers. Let’s take Machine Learning (ML) as an example:
Our customers want decision-relevant, ML-generated forecasts exactly when a decision is to be made or an action is to be initiated. In a way that integrates optimally into workflows. Hardly anyone cares what the Python code looks like or exactly which model is applied. It needs to feel as “normal” and intuitive as, say, facial recognition on a cell phone.
What does “Data Analytics as a Service” involve in the Digital Transformation Office (DTO) at thyssenkrupp Materials Services?
On the one hand, the so-called core, i.e. the algorithms. Typically, these were developed in a preliminary project on the basis of a defined data set and hypotheses from the specialist department. We call this phase Data Lab. As part of the service, the algorithms are continuously refined. Much more important, however, is the integration of new disciplines in service design. In addition to data engineers and data scientists, software architects and user experience designers (UX) are needed here. Because the focus here is: How do we make the results of the algorithms available to customers as efficiently as possible? Email, Microsoft Teams, apps, web services, messages, database entries – many things are possible here. The customer decides.
What advantages and potentials do data analytics and machine learning offer for operational business?
Strictly speaking, data analytics is old hat. Our colleagues have been doing good, data-driven analysis and improvement for decades. One example is Six-Sigma, a very standard and widespread method for data-driven process improvement that has been in use for over 20 years.
In my view, this has always been a key competitive advantage and remains so. What is fundamentally different today, however, are the technical capabilities of data analytics. In addition to cloud services and growing open-source libraries, e.g. for machine learning, this can be seen in the fact that data analytics is a discipline all of its own. So here, for example, particle physicists tell the operational logistics planner where which materials are best stored. Those who quickly overcome their skepticism in this area will leverage precisely the potential that remains after decades of operational improvement. For themselves and their customers.
What challenges are thyssenkrupp Materials Services and thyssenkrupp customers facing today that require the know-how of data analytics experts?
The world of tomorrow is also built on hardware. Cars, houses, cell phones, robots – materials are needed everywhere. In the process, supply chains are becoming more complex, as the products themselves are also becoming more complex and contain, for example, more and more components and/or more variants. This makes securing supply at globally competitive prices challenging.
The resulting optimization problems, such as finding good solutions in a very wide parameter space, are difficult. In addition, it is important to weigh up what is being optimized for, e.g. price, speed or risk. Data analytics offers decisive approaches to solving these problems, e.g. through simulations of very many scenarios or the recognition of complex patterns that are hidden from the human eye. In my eyes, modern supply chain solutions are not possible without data analytics.
What fascinates you about working in data analytics?
It is refreshingly factual, as it is purely data-driven, and there are always surprises.
I worked for a long time in Operational Excellence, focusing on series production. When the first machine-learning supported dashboard ran on a screen on the shop floor at one of our international sites, my heart sank.
Data analytics is involved in almost every major new technology today. What is the value of data analytics to industry and business today?
As I said before, data analytics has always helped us get better. However, depending on the industry and the market you’re in, mastering advanced methods like Big Data Analytics and Machine Learning is the key success factor today. But just not everywhere. And data analytics are not always the solution, sometimes the problem lies in much more trivial things.
How do you think the importance of data and data analytics will change in the future (and why)?
The importance is already very high today and will increase in my opinion.
The development in the area of data standards and platforms will be very exciting. Most companies are still reluctant to share their data. I can well understand that. Because the worry is that someone else will lift the treasure trove of data and you yourself will just become a data supplier. But with the necessary security and added value for each participant, a shared eco-system around data within a value chain could massively change the current structures. There is great potential behind this and I am curious to see what the next few years will bring in this area. Every company should prepare for digital supply chains and deal with data-driven business models.
Mr. Klemm, thank you very much for the interesting interview!