nflparser

Routines for parsing and analyzing NFL games.

Copyright (c) 2013 Andrew Werner

Free for public use so long as you give me some sort of credit, or at
least some notice, however obliquely or back-handedly. 
------------------------------------------------------------

This package includes a set of classes that should be of some use in 
interpreting NFL play-by-play data.  This data can be found for free
since the year 2002 at the Advanced NFL Stats website:

http://www.advancednflstats.com/2010/04/play-by-play-data.html

The following classes work well with the raw files.

--> the GameFactory class will take a file and return an iterable of games
--> the PlayMaker class will take a row and return a play having
    relevant information, including information from the description.
--> the parse_play and parse_plays routines parse the contents of the
    'description' column into a series of 'segments' that can be aggregated
    to provide useful information about persons involved with the play
    or otherwise interesting surrounding circumstances.

This framework differs from other framework in the level of detail it provides 
in regards to the 'description' column.  Compared with simpler techniques
based on token recognition and regular expression parsing, it is capable of 
recognizing complex play outcomes, such as successful and unsuccessful 
challenges, turnovers (and the outcome of the return), penalties (including
offsetting or superseded penalties), and other exotic situations such as
laterals.

Most of the action will occur with developing interesting PlayMaker classes
that deal constructively with the output of the parse_play routine.