Determining when baseball games become unique

counter.py

#!/usr/bin/env python

import csv
from itertools import groupby
from collections import Counter

reader = csv.DictReader(open('MLB.CWEVENTS'))
writer = csv.writer(open('sequences.csv', 'wb'))

it = groupby(reader, key=lambda row: row['GAME_ID'])
count = Counter()

for gid, events in it:
    print gid
    sequence = []
    all_events = list(events)

    # keep incrementing and re-running this script until
    # .most_common() returns all 1s.
    for event in all_events[:5]:
        sequence += list(event['PITCH_SEQ_TX']) + [event['EVENT_CD']]

    joined = ','.join(sequence)
    count.update([joined])
    writer.writerow([gid, joined])

# Grep for the output sequence in sequences.csv to figure out
# which GID produced it.
print count.most_common(3)

process.sh

#!/usr/bin/env bash

function run_cwevent {
  if [ ! -f $2 ]; then
    cwevent -n -f 0-10,7,29,34 -y $1 ${1}*.EV[AN] > $2
  else
    cwevent    -f 0-10,7,29,34 -y $1 ${1}*.EV[AN] >> $2
  fi
}

pushd 2000seve/
for year in $(seq 2000 2009); do
  run_cwevent $year ../MLB.CWEVENTS
done
popd

pushd 2010seve/
for year in $(seq 2010 2019); do
  run_cwevent $year ../MLB.CWEVENTS
done
popd