CaptainBash

@EliotJackson: Sounds understandable to me now. Maybe I'm to focused on these p-values, but that's how I learned proofing hypothesis. Besides, I just wanted to spit a little nerd talks as I was totaly in writing R-scripts for my studies this afternoon. Keep up the good work Eliot and James

@EliotJackson:Does that mean you've already tried boxplots and p-values? Maybe you would have to adjust the significance level for something like that. Normally it's p=0.05, maybe you'd have to find a lower level. And so we are again in the discussion that you have to think about what the definition of significance is. I wrote the note fast last night without seeing the used dataset. That was only my consideration how I would go about it, also concerning representation. During the night I tried to write a R-script myself. Unfortunately it already failed with the official UCI results to create the dataset. The calculation with the time format hh:mm:ss,000 has pissed me off too much.

@jamessmurthwaite: Great article and I think it's very good to get some statistics presented on the start page from time to time. However, I have a few comments about your statistics. (Nerd talk on): Why not just display the gaps per race and the total gaps per year, as a box plot, and have the significance calculated. Depending on the distribution of the data, you have to proceed in different ways in order calculate the exact P-value (proof of hypothesis). Either at the end by Analysis of Variance (ANOVA) and Tukey-HSD-test or Kruskal Wallis Test and Man-Whitney-U as post-hoc-test. By adding environmental parameters, it would also be possible to analyse everything multivariate, e.g. by non-metric-multidimensional-scaling and environmental fitting. The whole thing can then be displayed nicely as a graphic with vectors as drivers. Then you really get an answer to the question whether the gaps are getting smaller, and if so what is responsible for it (change to 29, rain, wind, track length, changed track to last year etc.) I'm sorry, but a statistical analysis without P-values is not transparent for me, and not particularly scientific. I think you can get more out of the data than qantile-quantile-norm plots and regression-lines. (Nerd talk off)

Well, and who won #not2bad face contest??

yeah it was the best decision to buy a second lyrik to replace my hold fork and not to repair my brakes at the car!!

Pro men: 1st Aaron Gwin 2nd Loic Bruni 3nd Troy Brosnan Pro women: 1st Manon Carpenter 2nd Rachel Atherton 3rd Emmeline Ragot