Data consultancy SANDY Energized Analytics believes in the importance of available data and how to utilize it for profitability. In this interview series, our three co-founders Claudius Hundt, Sebastian Scholz and Peter Karcher share with us how they discover the hidden treasure that lies within data. Data science was in large part developed along with the movement towards digitalization. Many innovations and optimization of existing processes are based on the use of data. In order to better understand, it is helpful to take a closer look at the possible applications of „big data” and how data scientists operate, which will be explained by Peter Karcher, Co-Founder und Head of Data Science at SANDY.


How long have you been with the SANDY Team?

I have been here from the start, even before SANDY was founded.


What possibilities do businesses have to create added value through data?

Generally speaking, data has great characteristics: it’s objective, fast and plentiful, permanently retrievable and cheap. As such, data can be used, for example, to make the best possible decisions, improve operating processes, or to introduce a variety of new features.


What are interesting possible applications of data predictions?

Predictions that are integrated into entirely automated processes are impressive. These have to be robustly built and intelligently designed so that they can be self-learning. Then data can unfold and prediction models can evolve to their full capability.


How does your company use data analytics?

We, the team behind SANDY, develop predictions and intelligent algorithms. Data analytics is an important step in that process. We use data analytics to understand individual problems in detail, so that we can build adequate predictions and models, which consider all insights and coherences and depict them.


What will be a daily application of data analytics and advanced prediction in your opinion?

Basically, everybody already knows applications of data analytics: self-driving cars or individualized advertisement online. In the future, applications that we as customers will hardly recognize will be found in more areas of our lives. For example, a smart shelf or warehouse that will know when it is empty (Internet of things/IoT) and orders automatically according to demand. Many things will not be noticed to the average consumer, but that does not mean it is “malicious data espionage;” rather, it will help optimize adjusting processes and make them simpler and better.


What distinguishes a good data scientist?

A data scientist needs three qualities, which are equally important: subject-matter knowledge, communication, and network.

Expert knowledge is the tool of the trade, which is acquired at university or through projects. Here the profile changed substantially over the last three years. The rising demand of data scientists is leading to a sharpened job profile.

Communication is important. The ability to be able to speak to a “problem owner” about his business, to really listen and understand their problem, to find a common language, and in the end translate the problem into a mathematical question is a recurring challenge. Likewise, results and insights have to be returned in a way, so the customer can make something of them and everybody can gain further insights.

Lastly, networking is always important. Maybe someone cannot solve a problem but knows a colleague that has solved a similar problem or knows of someone who has already solved it. One has to be ready to share personal knowledge and will in turn be given new insights.

The three qualities above are not entirely my own, but were introduced to me by my doctoral adviser. Back then I thought, “yeah, right!”, but really thought that expert knowledge was everything. But today, I have to say, he was right. Expert knowledge alone is not sufficient.


How does data analytics get better with Big Data?

For me the debate between Big Data and Small Data is the wrong focus. I don’t need Big Data to build good prediction models. Intuition – recognizing problems and understanding them – as well as a knowledgeable counterpart to help solve problems with data is key. When you cooperate well with customers, data analytics becomes really good. If you don’t, you will just receive average results.


What are some undiscovered possibilities of big data?

Currently everybody speaks about big data. But in my opinion, they speak too much about the “could” or “should” aspects. Too often we discover an underlying expectation of our clients: they see that Google has a lot of data and a well-functioning business model and associate the two. They conclude that if we simply collect a lot of data, we will have a well-functioning business model as well. And we “should” and “could” do the same. But in fact it is way more complex.

Often this “could” and “should” mentality even occurs in specific use case ideas. To start building something new from existing data is good and correct, but it is even better to start with an existing problem or a desired feature and ask yourself in a second step if data can help solving these problems. This is the rarer working model when we talk about Big Data. This might not create the one undiscovered possibility, but it is an important thought process with the vital potential for data-driven additional value.


How to recognize your work desk?

I envy people that have a clean and meticulous desk. I commit to do the same.