I’ve had a lot of students over the years, and I’ve realized that even those who’ve taken a probability and statistics class often lack a gut feeling intuition about how to deal with data. I’ve written a (first draft, admittedly) version of a “Probability for Science Students” (pdf).

Comments are welcome (especially for straight-up errors, although comments from fellow physicists who might suggest additional topics are welcome, too). A bit of warning, it is extremely technical, and is designed for undergraduates or grad students in physics or closely related disciplines. It is not even for mathematicians who might find my lack of rigor disturbing.

**-Dave**

This looks great Dave, something like that was missing during my undergraduate years. When I get a chance to go over it a bit more I’ll give some details. In the meantime,

1] The enjoy link is broken in the post

2] Spacing is wrong for operators e.g. Y_n mod n

3] Wait long enough and not _everything_ is a Gaussian. Central limit theorem only applies when the variance is defined and finite (e.g. try drawing from a Lorentz distribution).

4] Matplotlib is good, seaborn makes it beautiful. One line import for nicer plots: http://stanford.edu/~mwaskom/software/seaborn/

Useful topics that do not seem covered:

1] Generating random numbers pulled from any distribution via the CDF, you mention it but an explicit example would be better

2] Yule-Simpson paradox is worth knowing since you had a stock disease example in there anyways

3] Kolmogorovâ€“Smirnov test

Obligatory XKCD’s

http://xkcd.com/552/

http://xkcd.com/882/