

In [1]:

    
%load_ext load_style
%load_style talk.css
from talktools import website

conda and anaconda

anaconda

anaconda is a Scientific Python distribution, developed by , from their web-page:

Anaconda is a completely free enterprise-ready Python distribution for large-scale data processing, predictive analytics, and scientific computing

some advantages:

free
cross-platform
195+ of the most popular Python packages for science, math, engineering, data analysis, ... notably:
- NumPy
- SciPy
- Pandas
- IPython
- Matplotlib
- Numba
- Blaze
- Bokeh
easily install, update, revert Python packages (with dependencies ..)



In [2]:

    
website('https://store.continuum.io/cshop/anaconda/',width=1000)









    Out[2]:

conda

conda is a package and environment management system akin to virtualenv

example, I want to test the iris and cartopy packages.

These are developed by the UK Met. Office.

iris is a Python package for analysing and visualising meteorological and oceanographic data sets
cartopy is a Python package for advanced map generation with a simple matplotlib interface.

first create a conda environment named iris, with the minimum: python and ipython [+ libraries for the notebook]

ᐅ conda create -n iris python ipython ipython-notebook

then:

ᐅ source activate iris

and to install (in the iris environment)

ᐅ conda install -c https://conda.binstar.org/scitools cdat-lite
ᐅ conda install -c https://conda.binstar.org/scitools iris
ᐅ conda install -c https://conda.binstar.org/scitools cartopy

A (quick and dirty) exploration of iris and cartopy can be found in this notebook

IMPORTANT: you need to launch a new instance of the IPython notebook from the iris environment:

~/Documents/talks_seminars/metocean/notebooks  ᐅ source activate iris

(iris)~/Documents/talks_seminars/metocean/notebooks  ᐅ ipython notebook

binstar

binstar is a free service for hosting public packages for pip and conda

For example, the GDAL library is not available through Anaconda (the package distribution), but some people have compiled it on different platforms, packaged it according to the conda package management system specifications, and made it available through binstar, and it can be install via conda.



In [4]:

    
website('https://binstar.org/dashboard', width=1000, height=700)









    Out[4]:



In [3]:

    
import iris









    



---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-3-005053b25f37> in <module>()
----> 1 import iris

ImportError: No module named iris



In [3]:



In [ ]: