Playcount.py

playcount.py

import sys
import random
import math

if len(sys.argv) != 3: 
	print "Usage: ./playcount.py <file_to_parse> <output_counts_to_file>"
	exit(1)

countmap = {}

# count occurrences per user-video pair
handle = open(sys.argv[1], 'r')
for line in handle.readlines():
	parsedline = line.strip().split()
	key = (int(parsedline[1]), int(parsedline[2]))
	if key in countmap:
		countmap[key] += 1
	else:
		countmap[key] = 1

handle.close()

# create a testing set and let the remaining be the training set
samplemap = dict([(s, countmap.pop(s)) for s in random.sample(countmap, int(math.ceil(len(countmap)*0.25)))])

# store everything
def writefile(filename, data):
	outhandle = open(filename, 'w')
	for (user,video),value in data.iteritems():
		outhandle.write("%d %d %d\n" % (useridmap[user], videoidmap[video], value))
	outhandle.close()

writefile("%s-test" % sys.argv[2], samplemap)
writefile("%s-train" % sys.argv[2], countmap)

Jag okynnesrefaktorerade lite. :-D

Oj, missade din kommentar helt. Vad refaktorerade du sa du?

Edit: ah, såg din fork nu. Snyggt. Tack för tipset.