DutchDataBusters: matchmaking between students and businesses
How the successful start-up helps students get interesting assignments in the field of data analysisRead more
- Education , Research , Student
DutchDataBusters: matchmaking between students and businesses
There’s a growing demand from industry for academic knowledge in the area of data analysis, but the academic and business worlds are often too far apart to work together efficiently. DutchDataBusters brings the university and industry together by connecting Data Science and Computer Science students to businesses and providing them with guidance in solving concrete issues. This enables the students to put their academic experience into practice and build experience while helping businesses overcome their challenges: a win-win situation. One of the most interesting clients of the scheme to date: Real Estate at TU/e.
DutchDataBusters is an initiative by CodeNext21, a company that specializes in anomaly detection based on Machine Learning. It is housed on campus, in the Twinning building, which is a breeding ground for innovative businesses in the fields of Data Science, AI, Blockchain, ICT and software. Bram Cappers, one of the founders of CodeNext21 and initiators of DutchDataBusters, has a TU/e background himself: after completing his bachelor’s and master’s, he did a PhD on data visualization here. These days, he works at the Department of Mathematics and Computer Science as an assistant professor once a week.
Having one foot in business and another in academia, Cappers is excellently positioned to connect both worlds and enable them to work together better. This is exactly what he’s doing with DutchDataBusters, a successful start-up that helps students get interesting assignments from industry while helping businesses meet their data analysis challenges.
Formula 1 engine
It all started with the technology Cappers developed during his PhD. This technology became the foundation for ML21™, a platform used by CodeNext21 to look for deviations in data streams (aka ‘anomaly detection'). This involves analyzing large amounts of data to find suspicious or deviating data. “It may concern huge Excel tables of financial transactions or patient information,” Cappers says. “With all those data coming in, you have to determine very quickly if all of it’s okay. If, for instance, there’s something fraudulent happening or a patient is experiencing heart failure, you want to pick up on this asap.” The smart technology is capable of comparing large amounts of data and, based on this, indicating when something deviates or – as Cappers phrases it – ‘looks fishy’.
This way, you can analyze a very wide range of data and use the technology for different purposes. “It’s like a Formula 1 engine that can be used in lots of different cars,” says Cappers. “You can deploy it to detect fraud in the financial world and monitor patients in medical healthcare, but also in logistics, processes in manufacturing industry, sports…” Cappers continues to list possible things you can use data analysis for, which begs the question: what can’t you use it for? Cappers admits: “This technology’s broad applicability is its strength, but also poses a challenge. We’re constantly looking for new places and ways to use the technology for solving real-world problems.”
Another thing that makes this technology unique is that the system can tell you exactly why it thinks the data deviate. “Most algorithms simply notify you of something strange going on, but leave you in the dark as to why they drew this conclusion,” Cappers explains. “But if your bank account is blocked because of strange transactions, you want to know why. Our platform cannot only detect deviating data, but it also provides a clear explanation of why the data deviate. This allows you to communicate the results to your client in a clear and transparent manner, and solve the problem in a quicker and easier fashion.”
“Early on we noticed there’s a great need for this technology in industry, because businesses have all kinds of problems where anomaly detection would come in handy,” Cappers says. And the businesses often don’t have the required knowledge and time to solve these issues themselves. When Cappers noticed students at the university were looking for interesting graduation assignments, he realized there were two stakeholders that both had their own needs and that were a good match, but that were having trouble finding one another. “For students, it’s pretty hard to get an insight into what problems businesses have without professional experience and connections.” Cappers and his business partner Remco Eikhout, CEO of CodeNext21, thought this was a shame. But then they realized they were part of both the academic and the business world, which meant they could cater to both parties.
Cappers and Eikhout started investigating how they could bring the two worlds together and this sparked the idea of setting up DutchDataBusters, an initiative that matches students and businesses. This enables the students to work on interesting projects and build experience in the professional field, while helping businesses overcome their challenges by means of data analysis: a win-win situation. “What we do can be seen as matchmaking between students and businesses,” Cappers says. “This creates collaborations that are of interest to both parties.” As a non-profit organization, DutchDataBusters primarily aims to improve the ties between the university and industry. “But of course it’s also interesting for us to look through the eyes of the students to learn what’s going on at those businesses and acquire new insights that way,” Cappers admits.
Not in the trash
DutchDataBusters facilitates collaboration between students and businesses in several ways. Project supervisors help the students find a suitable assignment, taking into account their interests, such as healthcare or sports. Because data analysis can be used in almost all fields, Cappers reiterates. “We look for a regional business that dovetails with the student’s interests and assist in drafting a concrete problem definition. The latter has to be both interesting to the business and meet the criteria of a graduation subject.”
DutchDataBusters also offers students office space to work and assistance at the project level, for instance in overcoming obstacles or reporting to the client. “This way, we make sure that a student’s report doesn’t end up in the trash because it’s too technical,” Cappers says. “On the one hand we guarantee that the outcome is sufficiently concrete and relevant to the business case and, and on the other hand, that the student applies theoretical knowledge that is fundamental enough to be recognized from an academic perspective.” When it comes to academic aspects, students are under the guidance of an academic supervisor, who’s typically a full professor. And, not entirely unimportantly, students are paid for their work by the client. For businesses this is a very affordable solution and for students it’s a nice way to make a bit of money on the side, so once again it’s a win-win.
Another important task of DutchDataBusters is to safeguard the security of data. For understandable reasons, businesses are often hesitant to share their data. An initiative like DutchDataBusters, led by experts with a lot of experience in data analysis, can help win the client’s trust. Businesses are more likely to collaborate with students if there’s a trustworthy intermediary that guarantees the secure processing of their data.
Readied for reality
“All students around here are doing the same thing, but working on different issues,” Cappers says. During the sessions DutchDataBusters organizes, students give presentations on the problems they’re trying to solve. This allows other students to contribute ideas that may help move the project along. “One’s working on medical data, another on financial data. Those are basically unrelated, and yet there are similarities,” says Cappers. “From a data perspective it doesn’t matter if you’re working with patient records or bank transactions.” DutchDataBusters also organizes workshops, for instance on sharing scientific results with clients in a clear way.
During the project, which takes anywhere between six and nine months, students gain practical experience and learn a lot about how a business functions. They themselves are, for instance, the client’s first point of contact, which for most of them is a completely new experience. They also develop professional skills that are important in business, such as collaborating and presenting. “With us, students are being readied for reality,” says Cappers. “Once they’ve graduated, many of them will do the same they’re doing here. So thanks to this experience they’ll know what to expect after their studies.”
DutchDataBusters has already enabled dozens of bachelor’s and master’s students to find a paid traineeship or complete a graduation project, resulting in many innovative solutions in several branches of industry. One student, for instance, analyzed data on absenteeism and investigated how you can get people back to work more quickly. Another studied fraud in phone traffic and yet another student looked into the optimalization of goods transport using data analysis.
If you’re not allowed to exceed your current energy usage, but you do want to keep growing, you have to make smarter use of the energy
The most interesting assignment to date came from closer than expected, namely the university itself. Real Estate, which manages the TU/e buildings, came to DutchDataBusters with the question of how energy use on campus could be optimized. Due to the energy transition, increasing energy rates and the planned growth of the university, there’s a great need for smarter energy use on campus. “The university wants to double in size, but the power grid in the Netherlands is pretty full already,” Cappers explains. “If you’re not allowed to exceed your current energy usage, but you do want to keep growing, you have to make smarter use of the energy and look into better management of your equipment.”
By using a digital twin, a virtual representation of reality, you can map out the university’s energy usage. “This allows you to see how it would be impacted by certain changes, such as installing extra solar panels or charging stations, or how you could switch on and off certain equipment to avoid peaks in power usage,” Cappers explains.
What originally started as a student project, led to the founding of TIBO energy, a company that investigates how you can use smart algorithms for better management of energy usage in order to lower energy costs, reduce CO₂ emissions or keep growing sustainably. Cappers and the rest of the team are currently developing an energy platform that could play a key role in the future growth of the university. This is how academic knowledge can result in an innovative tech solution that benefits the university itself, bringing things full circle.
Not all students are admitted to the scheme, as the project supervisors at DutchDataBusters carry out an advance selection procedure. When asked what’s most important for a student to have, Cappers reacts quickly and without hesitation: “Motivation. If you’re motivated, you stand the biggest chance of making the project a success. It’s easy to get started, but you really have to show determination. When you’re enjoying yourself, learning comes naturally. And you receive guidance, of course.”
Are you looking for a traineeship or graduation project in the area of data analysis? Don’t hesitate to sign up at DutchDataBusters.