Spiegel Mining: Data exploration with David Kriesel
Created:
You can see for yourself how Kriesel uses data exploration methods to discover some humorous, surprising and even disquieting findings about politics and Germany: Watch the recordung of our Online Seminar.
I asked David a few questions about his Spiegel Mining project.
How did you get the idea to analyse Spiegel Online articles?
It was a mixture of technical and social curiosity. When I learn new techniques, I usually pursue a "learning by doing" approach. This means, I create some kind of toy project that forces me to use the things I want to learn. I am interested in media in general, so one of these toy projects was downloading Spiegel Online.
Were you surprised to get so many hits on your YouTube video? Why do you think it resonated so well?
Chaos Communication Congress, the conference where I gave the first talk on the topic, is one of the largest hacker conferences in the world. If you manage to get a slot there, be sure to prepare for a bit of load on your servers. :-) But to be honest, I was gobsmacked at how much resonance came out of it. It took me more than two months to answer the emails I got; some are still unanswered today.
On the other hand, it is not so easy to tell what makes a talk on a nerdy topic go viral. I think nowadays, everybody is talking about data analysis in one or the other way. So besides learning new things, a side purpose of the toy project was to make plausible for non-IT-professionals to see what today's data analysis is. In easy, understandable words, without tech stuff, without formulas and -- maybe most important -- on a data set they all know, because almost everybody reads Spiegel Online at least once in a while. I think this mixture just fit in.
Data is everywhere. Why is it so important to analyse it? What is the value for smaller companies who might not have a statistician or data scientist on board?
Data indeed is everywhere, but in many cases it's just lying around unused. It's basically a worthless asset. It even makes you pay for storage. But if you analyse data, you can sometimes, at surprisingly low cost, convert your asset into knowledge, and turn an expense factor into something valuable. This is independent from the size of the particular company. I know many small companies using data science even if they don't have a dedicated person on board. Sometimes hiring a freelancer for one or two days already helps a lot. The barriers are lower than most companies think. For standard analyses, there's analysis software. You don't even need to be able to code any more.
What makes a good data visualisation?
A good visualisation in my opinion enables people with domain knowledge (i.e., knowledge in the specific domain the data you analyse comes from) to find patterns in the data intuitively, without the further help of a data scientist. It empowers them to draw the right conclusions, and if you did well, it's also beautiful, so people want to look at it again and again (which makes them draw more conclusions). There is only one high-bandwidth connection to the brain, and that's the eyes.
You have other interests outside of data science. Tell us about your ant colony.
Sure! For a long time, I have been interested in swarm behavior, and this is why I have kept ants for quite a while. Many people are not aware that you can actually keep ants as pets (no obligation to name them, though). You build them a small terrarium and then watch them build their nest, grow eggs, collect food, carry out the garbage, and so on. It’s quite interesting and also relaxing to watch them do their logistics and build their streets. As a plus, ants don't get angry when you go to vacation; they are fine for quite a long time without being fed, they don't smell bad, and they don’t make any noise. :-)
If you'd like to hear more from David watch the recording of his online seminar.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.