Tour de France Wrap-Up

The final installment of the 2011 Tour de France graphs, and what a fun race that was this year! Plenty of excitement (for some not so fun reasons) early on and then plenty of exciting racing towards the end. Good stuff…

As usual, you can click on them for slightly larger versions. Here’s the same graph for 2011 broken down by team: Read more

Tour de France Graphs

Rest day number two is upon us, so it’s time to update my graphs tracking the race thus far. First the standard ‘bump chart’ showing the GC picture through the first 15 stages:

And for comparison, the same graphs for all the Tour back to 2005 (click for larger version): Read more

Giro d’Italia: Stage 21

Final Giro post! First we have the final picture for the whole race:

and a higher quality version here. And then we have the breakdown by teams:

And the matching high quality version here. Finally, some higher quality graphs looking at some folks who competed in both this and last year’s Giro and who did particularly better or worse than last year.

Giro d’Italia: Rest Day 2

We’re through Stage 15 of the Giro, so here are some updated versions of the accompanying graphs. I made these “high quality” (i.e. PDFs) in case anyone wants to repost them elsewhere. Here’s the overall GC picture:

And then we have the same data broken down by team: Read more

New Biathlon Individual Graphs

Biathlon’s cool in part because there’s so much more data to play with. One seemingly minor thing I’ve been struggling with is how best to graph an athlete’s overall race results and their shooting statistics on the same graph.

The obvious answer of just overlaying them and including a second y axis (since shooting accuracy and percent back are on very different scales) is a big no-no in the statistical graphics world. Dual y-axis graphs are very bad news, almost as bad as the hated pie graphs. In fact, they are so bad that the statistical graphing package I use literally can’t do them!

It took a little tinkering, but I think I like this solution just fine. Basically, I just put the two time series (percent back, shooting accuracy) in vertically grouped panels for each skier. The one drawback (so far) is that this means I can’t only extend this to multiple athletes by making the graph wider. It’s probably possible for me to “wrap” this around, but, well, you don’t care about the details. It’s complicated.

Anyway, as an example here are the graphs for the top five finishers from each of yesterday’s mass start World Cup races in Fort Kent. They’re big, so click on them for the full versions:

The top panels are actually percent back from the median skier (rather than from the winner) which I’ve learned to prefer. The blue line tracks the median result over time. Lower values are better. The bottom panels are shooting accuracy for each season, where higher values are obviously better.

One important piece of information that I decided not to squeeze on here is the number of shots, which early on for each skier can be very low. So, beware small sample size.