Python + R workflow to build a graph of a Kaggle competition leaderboard.
See datascience_leaderboard.png
for an example.
The python script leaderGrabber.py
downloads the Kaggle raw leaderboard data
described in config.py
and saves the CSV out to disk. It then calls the R
script makeGraph.R
, which must be configured with some basic parameters (score
range, min/max date, file names, title), which creates the graph and saves a png.
leaderGrabber
then uses sips to resize the image, if there are any sizes
defined in config, for use as thumbnails, etc. It then scps it to a destination
of your choice.
As a somewhat adhoc system this is a bit ugly. It uses Python and R, and seems to require a reasonably modern R (I've only tested it on Python 2.7 and R 2.15 on OS X).
The only non-standard dependency is that python uses Mechanize to automatically download the CSV, because Kaggle wanted me to login.