The Hitchhiker's Guide to Python

Chapter 10. Data Manipulation

This chapter summarizes the popular Python libraries related to data manipulation: numeric, text, images, and audio. Almost all of the libraries described here serve a unique purpose, so this chapter’s goal is to describe these libraries, not compare them. Unless noted, all of them can be installed directly from PyPI using pip:

$ pip install library

Table 10-1 briefly describes these libraries.

Table 10-1. Data tools

Python library	License	Reason to use
IPython	Apache 2.0 license	Provides enhanced Python interpreter, with input history, integrated debugger, and graphics and plots in-terminal (with the Qt-enabled version).
Numpy	BSD 3-clause license	Provides multidimensional arrays and linear algebra tools, optimized for speed.
SciPy	BSD license	Provides functions and utilities related to engineering and science, from linear algebra to signal processing, integration, root finding, statistical distributions, and other topics.
Matplotlib	BSD license	Provides scientific plotting.
Pandas	BSD license	Provides series and DataFrame objects that can be sorted, merged, grouped, aggregated, indexed, windowed, and subset—a lot like an R Data Frame or the contents of a SQL query.
Scikit-Learn	BSD 3-clause license	Provides machine learning algorithms, including dimensionality reduction classification, regression, clustering, model selection, imputing missing data, and preprocessing.
Rpy2	GPLv2 license