Let me make things much easier on you - visit Christoph Gohlke's Python Extension Packages for Windows Repository, and get every single package you need from there. NumPy and SciPy are compiled using Intel's Math Kernel Library, which makes certain functions significantly faster. All the supporting libraries are either contained in the installer, or linked so you can easily get an installer for the version of Python you're using. And, most importantly, any dependencies are listed, so for example when you go to the pytables
link you'll see that it requires numexpr
, and that in turn requires NumPy
.
I'd also recommend getting pandas
and matplotlib
, along with their recommended dependencies, as well as IPython
, which makes working with scientific and visual data a lot easier, and it makes iterative development in general much more fun than with the command line and/or IDLE alone with features like command completion and inline documentation hooks.