This repo contains the slides, data, and Jupyter Notebook for the PyData Seattle 2015 talk Python Data Bikeshed. The goal of the talk is to help answer the following question:
I have data. It’s July 2015. I want to group things. or count things. or average things. or add things.
What Python library do I use?
This talk discusses the following libraries in some depth:
- Toolz: https://github.com/pytoolz/toolz
- Pandas: https://github.com/pydata/pandas
- Blaze: https://github.com/ContinuumIO/blaze
- xray: https://github.com/xray/xray
- bcolz: https://github.com/Blosc/bcolz
- Dask: https://github.com/ContinuumIO/dask
The following libraries or projects are briefly mentioned:
- Cython: https://github.com/cython/cython
- Numexpr: https://github.com/pydata/numexpr
- Numba: https://github.com/numba/numba
- Numpy: https://github.com/numpy/numpy
- Spark: http://spark.apache.org/
- Bolt: http://bolt-project.org/
- Dato SArray: https://dato.com/products/create/docs/generated/graphlab.SArray.html
- Dato SFrame: https://dato.com/products/create/docs/generated/graphlab.SFrame.html#graphlab.SFrame
- DistArray: https://github.com/enthought/distarray
- Biggus: https://github.com/SciTools/biggus
- Spartan: https://github.com/spartan-array/spartan
- Ibis: https://github.com/cloudera/ibis
- scikit-learn: https://github.com/scikit-learn/scikit-learn
- statsmodels: https://github.com/statsmodels/statsmodels/
- Bokeh: https://github.com/bokeh/bokeh
- seaborn: https://github.com/mwaskom/seaborn
There is an example Notebook that goes into much more detail on the core libraries discussed in the talk, with plenty of examples; it can be read directly in Github, or on nbviewer.
If you would like to run the example yourself, you will need to either pip or conda install the follow dependencies:
Babel==1.3
Cython==0.22.1
Jinja2==2.7.3
MarkupSafe==0.23
Pygments==2.0.2
SQLAlchemy==1.0.7
Sphinx==1.3.1
alabaster==0.7.6
backports.ssl-match-hostname==3.4.0.2
bcolz==0.10.0
blaze==0.8.2
certifi==2015.04.28
dask==0.6.0
datashape==0.4.6
decorator==3.4.2
dill==0.2.3
docutils==0.12
functools32==3.2.3-2
gnureadline==6.3.3
ipython==3.2.1
jsonschema==2.5.1
mistune==0.7
multipledispatch==0.4.8
networkx==1.9.1
nose==1.3.7
numexpr==2.4.3
numpy==1.9.2
numpydoc==0.5
odo==0.3.3
pandas==0.16.2
psutil==3.1.1
psycopg2==2.6.1
ptyprocess==0.5
python-dateutil==2.4.2
pytz==2015.4
pyzmq==14.7.0
requests==2.7.0
six==1.9.0
snowballstemmer==1.2.0
sphinx-rtd-theme==0.1.8
terminado==0.5
toolz==0.7.2
tornado==4.2.1
wsgiref==0.1.2
xray==0.5.2
The Blaze demo also requires installing Postgres, creating a pydata database, and populating a diamonds table with the correct schema. Postgres can be reliably installed with homebrew or conda, and the table created with the following SQL:
CREATE TABLE diamonds (
carat float,
cut varchar(255),
color varchar(255),
clarity varchar(255),
depth float,
"table" float,
price integer,
x float,
y float,
z float
);