-
Notifications
You must be signed in to change notification settings - Fork 11
Installing the scientific Python stack
Installing Python and the libraries used in this workshop from scratch is possible, but challenging and time consuming. The easiest way to get up and running quickly with Python is to either use a cloud service, or to install a bundled distribution of the scientific Python stack. Here are some Python distributions that are suitable for data analytic work:
-
Enthought Canopy (https://www.enthought.com/products/canopy)
-
Continuum Anaconda (http://continuum.io/downloads)
-
Python-xy (https://code.google.com/p/pythonxy/) -- Windows only
-
Pyzo (http://www.pyzo.org)
If you use Linux or MacOS, you can install core Python by first downloading the source tarball (pythonxxx.tar.gz) from (https://www.python.org/downloads), then following these steps (change pythonxxx to the specific file name):
tar --xz -xvf pythonxxx.tar.xz
cd pythonxxx
./configure
make
make install
Note that you need to have root access on the machine to do this. If you do not have root access, it is possible to install everything in an arbitrary location by using
./configure --prefix=/path/to/location
If you install Python in a non-standard location you will need to use the full path to the executable to launch it (or make an alias in your .bashrc or other shell configuration file).
Next you need to install each of the core libraries. Briefly, download an archive file for each library, e.g. numpyxxx.tar.gz, then
gunzip numpyxxx.tar.gz
tar -xvf numpyxxx.tar
cd numpyxxx
python setup.py install
Note that the gunzip and tar steps may differ depending on the archive format (.tar.gz, .tar.bz2, .tar.xz, .zip). The python command in step 4 should invoke whichever Python installation on your system you want to link to the libraries (i.e. if you have installed several Pythons, you need to use the right one here).
An alternative way to install the libraries (but not Python itself) is to use pip (https://pypi.python.org/pypi/pip). We won't provide details on this here, see the pip web page for instructions.
If you use a Linux distribution like Ubuntu, you can use your package manager (yum, apt-get, etc.) to download Python (which should already be present), and the various libraries (which generally are not part of the Linux distribution).
If you use Windows and have administrator access to your system, you can scour the web for the core language and various libraries in self-extracting executable (*.exe) format, and install them by running them (double click on the *.exe).
A good resource for up-to-date windows packages for Python is: http://www.lfd.uci.edu/~gohlke/pythonlibs