def generateKaggleSubmission(weights, outfilename):
with open(outfilename, 'w') as f:
writer = csv.DictWriter(f, fieldnames=['Id', 'Prediction'])
writer.writeheader()
# Test data is used for private leaderboard
testData = dataIterator(TESTKEY, test_mode=True)
for i, (counts, _) in enumerate(testData):
predictedLabel, _ = predict(counts, weights, ALL_LABELS)
predictedIndex = ALL_LABELS.index(predictedLabel)
writer.writerow({
'Id': 'test-{}'.format(i),
'Prediction': predictedIndex})
# Dev data is used for public leaderboard
devData = dataIterator(DEVKEY, test_mode=False)
devCorrect = 0
devTotal = 0
for i, (counts, label) in enumerate(devData):
devTotal += 1
predictedLabel, _ = predict(counts, weights, ALL_LABELS)
devCorrect += (predictedLabel == label)
predictedIndex = ALL_LABELS.index(predictedLabel)
writer.writerow({
'Id': 'dev-{}'.format(i),
'Prediction': predictedIndex})
devAccuracy = float(devCorrect) / devTotal
print 'Dev accuracy is ', devAccuracy, '({} correct of {})'.format(devCorrect, devTotal)
print 'Kaggle submission saved to', outfilename, ('. Sanity check: '
'public leaderboard accuracy should be '), devAccuracy, 'on submission.'
def generateKaggleSubmission(weights,outfilename):
with open(outfilename, 'w') as f:
writer = csv.DictWriter(f, fieldnames=['Id', 'Prediction'])
writer.writeheader()
# Test data is used for private leaderboard
testData = dataIterator(TESTKEY,test_mode=True)
for i,(counts,_) in enumerate(testData):
predictedLabel,_ = predict(counts,weights,ALL_LABELS)
predictedIndex = ALL_LABELS.index(predictedLabel)
writer.writerow({
'Id': 'test-{}'.format(i),
'Prediction': predictedIndex})
# Dev data is used for public leaderboard
devData = dataIterator(DEVKEY,test_mode=False)
devCorrect = 0
devTotal = 0
for i,(counts,label) in enumerate(devData):
devTotal += 1
predictedLabel,_ = predict(counts,weights,ALL_LABELS)
devCorrect += (predictedLabel == label)
predictedIndex = ALL_LABELS.index(predictedLabel)
writer.writerow({
'Id': 'dev-{}'.format(i),
'Prediction': predictedIndex})
devAccuracy = float(devCorrect) / devTotal
print 'Dev accuracy is ', devAccuracy, '({} correct of {})'.format(devCorrect, devTotal)
print 'Kaggle submission saved to', outfilename, ('. Sanity check: '
'public leaderboard accuracy should be '), devAccuracy, 'on submission.'
def setup_module():
# Need to do this because the dataIterator function depends
# on the BOW file to be generated.
global ac_train
global ac_dev
docsToBOWs(TRAINKEY)
docsToBOWs(DEVKEY)
ac_train = getAllCounts(dataIterator(TRAINKEY))
ac_dev = getAllCounts(dataIterator(DEVKEY))
def setup_module():
# Need to do this because the dataIterator function depends
# on the BOW file to be generated.
global ac_train
global ac_dev
docsToBOWs(TRAINKEY)
docsToBOWs(DEVKEY)
ac_train = getAllCounts (dataIterator (TRAINKEY))
ac_dev = getAllCounts (dataIterator (DEVKEY))
def evalClassifier(weights, outfilename, testfile, test_mode=False):
with open(outfilename, 'w') as outfile:
for counts, label in dataIterator(testfile, test_mode): # iterate through eval set
print >> outfile, predict(counts, weights, ALL_LABELS)[0] # print prediction to file
if test_mode:
return
else:
return gtnlplib.scorer.getConfusion(testfile, outfilename) # run the scorer on the prediction file
def evalClassifier(weights,outfilename,testfile,test_mode=False):
with open(outfilename,'w') as outfile:
for counts,label in dataIterator(testfile,test_mode): #iterate through eval set
print >>outfile, predict(counts,weights,ALL_LABELS)[0] #print prediction to file
if test_mode:
return
else:
return gtnlplib.scorer.getConfusion(testfile,outfilename) #run the scorer on the prediction file