All of this is open source, so if you want to try tesseract on your own data, I recommend to fork the original example. If you should be interested in my code for converting the Zugmonitor data into CSV, just contact me on Twitter and I will publish it as well.
The coordinated visualizations below (built with D3) show nearly 60,000 train connections from early 2012 (data courtesy of OpenDataCity). The dataset is 2.5MB, so it might take a few seconds to download. Click and drag on any chart to filter by the associated dimension. The table beneath shows the eighty most recent trains that match the current filters; these are the details on demand, anecdotal evidence you can use to weigh different hypotheses.
Some interesting views to explore: How does time-of-day correlate with arrival delay? Are longer or shorter train rides more likely to arrive early? How do train delay patterns differ between mornings and nights?