kgschart is a python pakcage for parsing KGS rank graphs into numeric data. Visit this page for the overview of the package.

• R version of the same functionality has been developed
• A web app has been published

• Python 2.7+ or 3.4+
• numpy pillow scipy pandas scikit-learn matplotlib

The installation is the easiest with anaconda/miniconda python distribution, since it simplifies the setup process for scientific computation libraries such as numpy, scipy, and scikit-learn.

If you use python distribution based on anaconda or miniconda based environment, first, install required packages by conda command:

$ conda install numpy pillow scipy pandas scikit-learn matplotlib pip

Then, install kgschart package by:

$ git clone --depth 1 https://github.com/kota7/kgschart.git
$ pip install --no-deps ./kgschart

Note that we should use --no-deps flag since required packages are already installed by conda.

Alternatively, download the package directly from GitHub

$ pip install --no-deps git+https://github.com/kota7/kgschart

The kgschart package works also on official (non-conda) Python (provided that dependencies are installed properly).

The following command tries to install the package along with the dependencies.

$ git clone --depth 1 https://github.com/kota7/kgschart.git
$ pip install ./kgschart

Alternatively, download the package directly from GitHub

$ pip install git+https://github.com/kota7/kgschart

If the installation is successful, following commands should run with no error.

>>> from kgschart import KgsChart
>>> from pkg_resources import resource_stream
>>> with resource_stream('kgschart', 'example/leela-ja_JP.png') as f:
....    k = KgsChart(f)
>>> k.parse()
>>> print(k.data.head())
#                        time      rate
#0 2016-03-21 22:19:01.165048  1.762470
#1 2016-03-22 13:51:03.495146  1.762470
#2 2016-03-23 05:23:05.825242  1.776722
#3 2016-03-23 20:55:08.155340  2.040380
#4 2016-03-24 12:27:10.485436  2.232779

See this page for more about the usage of the package.