Overview of libraries

Python libraries

Python itself is not very useful for scientific work. However by using Python together with some powerful libraries, many things become possible.

The "scientific Python stack" is a loosely coupled association of the core Python interpreter and a collection of powerful libraries. This is in contrast to, say, R and Matlab, where the linear algebra and graphics capabilities are "baked in" to the core language.

An advantage of the loosely coupled approach is that it is possible, as needed, to experiment with alternative libraries, even for fundamental things like array processing. A disadvantage of this approach is that sometimes the Python syntax does not allow data structures to be manipulated in the most natural way.

Here are the libraries we will discuss in this workshop:

• Pandas (http://pandas.pydata.org): data management

• Matplotlib (http://matplotlib.org): graphics and plotting

• Statsmodels (http://statsmodels.sourceforge.net): statistical models and data analysis tools

• Numpy (http://www.numpy.org): array processing and linear algebra, front end to a subset of Lapack, provides matrix algebra functionality similar to Matlab and R (but with some different syntax).

• Scipy (http://www.scipy.org): numerical routines such as integration, optimization, and special functions