Student Projects in Data Analytics for Industries at IMACM / University of Wuppertal
During this winter term, 23 undergrad students of mathematics and economath make a special experience. In teams with up to fife students, they are invited to regional industry facilities where they are confronted with a data analytics task derived from a concrete industrial need. The industrial partners stem from markets as diverse as energy, insurance, production and retail.
In the first round, experts from industry present their problem and the associated data (often up to several million of cases) to the students. Afterwards, a phase of data input, merging, cleaning and visualization follows. Here the students keep close contact with their problem sponsors, clarifying, e.g., the nature of outliers.
After the data is set, the modeling and analysis phase follows. The down selection of adequate statistical models and understanding of the mathematical foundations is a critical step. Using the R language, the statistical models are fitted to the data and are post processed using co-variable selection procedures from Akaike information to LASSO.
Finally, the results are visualized and put into industry compatible slides, leaving mathematical detail for the back-up section. The final presentation then is the climax of the seminar. However, in many cases both partners keep on working together through internships or theses.
This format is repeated every two years and at present is in its fourth round with fife industrial partners participating. Since its first start, more than a dozen enterprises joint the experience, many of them repeatedly.
Head of IMACM, Bergische Universität Wuppertal