Tips on using Pandas

Read Python for Data Analysis

Python for Data Analysis is very good as both a tutorial and reference. It also covers using iPython.

Use Pandas

For many things it seems initially that Python alone will suffice. It probably will. However, using Pandas gives you a lot of well implemented functionality for free and is a good choice for open ended projects.

Functional not OOP

Try to write functional (functions accepting functions as arguments) code rather than complex objects. For flexible, interactive analysis this is usually the way to go.

Don't just use Pandas

The visualisation (Matplotlib) is way behind R's ggplot2. Reluctantly I use it. Things seem to be improving for Python, e.g. Bokeh. If you are using Matplotlib, see Damon's guide.

Use Continuum.io's Anaconda distribution of Python

Some Python libraries are tricky to compile, this does it for you: Anaconda. The basic version is fine for most uses, but even the more advanced one is free for Academic users. This allows you to use things like Numba (roughly a Python JIT) which seems to be impossible to compile on Mac.