The pandas library is part of the Python language, so we can now proceed to install pandas. At the time of writing this book, the latest stable version of pandas available is version 0.12. The various dependencies along with the associated download locations are as follows:
|
Package |
Required |
Description |
Download location |
|---|---|---|---|
|
|
Required |
NumPy library for numerical operations | |
|
|
Required |
Date manipulation and utility library | |
|
|
Required |
Time zone support | |
|
|
Optional, recommended |
Speeding up of numerical operations | |
|
|
Optional, recommended |
Performance-related | |
|
|
Optional, recommended |
C-extensions for Python used for optimization | |
|
|
Optional, recommended |
Scientific toolset for Python | |
|
|
Optional |
Library for HDF5-based storage | |
|
|
Optional, recommended |
Matlab-like Python plotting library | |
|
|
Optional |
Statistics module for Python | |
|
|
Optional |
Library to read/write Excel files | |
|
|
Optional |
Libraries to read/write Excel files | |
|
|
Optional |
Library to access Amazon S3 | |
|
|
Optional |
Libraries needed for the read_html() function to work | |
|
|
Optional |
Library for parsing HTML | |
|
|
Optional |
Installing pandas is fairly straightforward for popular flavors of Linux. First, make sure that the Python .dev files are installed. If not, then install them as explained in the following section.
For the Red Hat environment, run the following command:
yum install python-dev
Now, I will show you how to install pandas.
For installing pandas in the Ubuntu/Debian environment, run the following command:
sudo apt-get install python-pandas
Install Python-pandas via YaST Software Management or use the following command:
sudo zypper install python-pandas
Sometimes, additional dependencies may be needed for the preceding installation, particularly in the case of Fedora. In this case, you can try installing additional dependences:
sudo yum install gcc-gfortran gcc44-gfortran libgfortran lapack blas python-devel sudo python-pip install numpy
There are a variety of ways to install pandas on Mac OS X. They are explained in the following sections.
The pandas have a few dependencies for it to work properly, some are required and the others are optional, although needed for certain desirable features to work properly. This installs all the required dependencies:
- Install the
easy_installprogram:wget http://python-distribute.org/distribute_setup.pysudo python distribute_setup.py - Install Cython
sudo easy_install -U Cython - You can then install from the source code as follows:
git clone git://github.com/pydata/pandas.git cd pandas sudo python setup.py install
The following methods describe the installation in the Windows environment.
Make sure that numpy, python-dateutil, and pytz are installed first. The following commands need to be run for each of these modules:
- For python-dateutil:
C:Python27Scriptspip install python-dateutil - For pytz:
C:Python27Scriptspip install pytz
Install from the binary download, and run the binary for your version of Windows from https://pypi.python.org/pypi/pandas. For example, if your processor is an AMD64, you can download and install pandas by using the following commands:
- Download the following file: (applies to pandas 0.16)
pandas-0.16.1-cp26-none-win_amd64.whl (md5)
- Install the downloaded file via pip:
pip install pandas-0.16.1-cp26-none-win_amd64.whl
To test the install, run Python and type the following on the command prompt:
import pandas
If it returns with no errors then the installation was successful.
The steps here explain the installation completely:
- Install the
MinGWcompiler by following the instructions in the documentation titled Appendix: Installing MinGW on Windows at http://docs.cython.org/src/tutorial/appendix.html. - Make sure that the
MingWbinary location is added to thePATHvariable, that hasC:MingWinappended to it. - Install
CythonandNumpy.Numpycan be downloaded and installed from http://www.lfd.uci.edu/~gohlke/pythonlibs/#numpy.Cythoncan be downloaded and installed from http://www.lfd.uci.edu/~gohlke/pythonlibs/#cython
The steps to install Cython are as follows:
- Installation via Pip:
C:Python27Scriptspip install Cython
- Direct Download:
- Download and install the pandas source from GitHub: http://github.com/pydata/pandas.
- You can simply download and extract the zip file to a suitable folder.
- Change to the folder containing the pandas download
to C:python27pythonand runsetup.pyinstall. - Sometimes, you may obtain the following error when running
setup.py:distutils.errors.DistutilsError: Setup script exited with error: Unable to find vcvarsall.bat
This may have to do with not properly specifying mingw as the compiler. Check that you have followed all the steps again.
Interactive Python (IPython) is a tool that is very useful for using Python for data analysis, and a brief description of the installation steps is provided here. IPython provides an interactive environment that is much more useful than the standard Python prompt. Its features include the following:
- Tab completion to help the user do data exploration.
- Comprehensive Help functionality using
object_name?to print details about objects. - Magic functions that enable the user to run operating system commands within IPython, and run a Python script and load its data into the IPython environment by using the
%runmagic command. - History functionality via the
_,__, and__variables, the%historyand other magic functions, and the up and down arrow keys.
Note
For more information, see the documentation at http://bit.ly/1Is4zIW.
IPython Notebook is the web-enabled version of IPython. It enables the user to combine code, numerical computation, and display graphics and rich media in a single document, the notebook. Notebooks can be shared with colleagues and converted to the HTML/PDF formats. For more information, refer to the documentation titled The IPython Notebook at http://ipython.org/notebook.html. Here is an illustration:
The preceding image of PYMC Pandas Example is taken from http://healthyalgorithms.files.wordpress.com/2012/01/pymc-pandas-example.png.
