Just finished Udacity's Statistics 101 and it was was a blast! Next up, Machine Learning (again). I'd recommend Stats before ML since ML really builds on the Stats formulas. Stat builds up to Linear Regression and Confidence Intervals. ML covers LR briefly then moves on quickly to more advanced mulch-variate analysis.
As an aside, I did most of the calculation work for this class in Python, then implemented the same algorithms in HP-48 RPL, then, near the end, started using R. There is a great R IDE here. http://flowingdata.com/2011/03/02/rstudio-a-new-ide-for-r-that-makes-coding-easier/
The base R package (you'll need both) is http://cran.r-project.org/
Looking back, I should have started with R. It makes the work of solving ML and Statistics problems so much easier.