Crazy for Data Analytics, or “Nerdy and proud!”
I discovered my love for data analytics some time ago. Yet it continues to scale new heights and, since the beginning of the year, it has got so far that I use the Tool Life Cycle to analyze virtually everything I have done and when, where and for how long I’ve done it. What can I say? I love data analytics!
That’s why I was really excited about the Porsche Hackathon in Stuttgart, where computer scientists, mathematicians and other data nerds tackled a Big Data Challenge.
48 hours, exclusive data provided from more than 4,500 vehicle test days, 26 people, 7 teams — that was the event in dry data. But what this fails to reveal on first glance is how educational the hackathon actually was. It was exciting to see how the teams structured and analyzed the vehicle data — and I came away with three important insights for my work at Porsche Digital Lab:
A problem looks totally different when considered from above, as opposed to from the side.
1. Data is art: Data visualizations are extremely valuable for gaining a better understanding of data bases and new perspectives. Actually, they only transform complex data into a world of images, translating measurements and statistics into colors, patterns and shapes. And yet, what emerges is so much more than a straight visualization. At the hackathon, the team “Die FIXen” visualized the data in a beautiful way, as you can see in the header image of this article — to me, that’s art. Art that can take on all kinds of different forms, as the students from Potsdam University of Applied Sciences demonstrated: They visualized the findings of the Bertelsmann Foundation’s Social Justice Index as an interactive installation. They used a number of different “sand molds” to illustrate the state of child poverty in 28 European countries. Presenting the study’s findings on a different level makes them easier to take in and understand.
2. It’s all a question of perspective: Seven teams have at least 26 different perspectives on one and the same issue — as a result, all the participants are stimulated to take new approaches. A problem looks totally different when considered from above, as opposed to from the side. At the same time, there are no right or wrong answers — only different viewpoints.
I am curious to see what perspectives will become a reality in the future thanks to new technologies and interfaces. And what viewpoint Artificial Intelligence will adopt in data analysis — and what visualization and interpretation it will choose to prepare its findings for human consumption.
And yet, what emerges is so much more than a straight visualization.
3. Trees help to classify data: I also learned a lot about classifying data, i.e. how to structure and organize data in order to make the findings concealed within them more accessible. This includes, for example, boosting, where several weak classifiers are automatically merged into a good classifier. And the Random Forest procedure — a very democratic way of classifying data. A “random forest” comprises different decision trees which each make a decision, each casting a vote in a grass-roots democratic way (if you’ll pardon the mixed metaphors). And the category with the most votes wins. This means: It then decides the final classification of the data.
Ok, ok, call me a nerd if you want.
But to those who claim that this is only exciting for techies, I say: Even the Museum of Modern Art has hosted an exhibition featuring the findings of data visualizations. And decision trees are used in countless settings — they are, among other things, a great tool for more complex personal decision-making in both your private and your working life.
If you love data analytics as much as I do, or would like to find out more, simply follow Porsche Digital Lab in Berlin & me on Twitter and be the first to hear about the latest projects and events.
