def sentiment_test(lt, nb, directory, output):
try:
f = open(output, 'w')
except IOError:
print 'Failed to open output file'
sys.exit()
f.write('<html><body><table border="1" cellpadding="2"><tbody>')
f.write('<tr><td>Run</td>')
f.write('<td>False Positive Rate (%)</td>')
f.write('<td>False Negative Rate (%)</td>')
f.write('<td>Accuracy (%)</td></tr>')
run = 1
fp = []
fn = []
acc = []
while True:
classifier = PostClassifier()
train_glob = directory + str(run) + '_tr*'
print 'train_glob: ' + train_glob
names = glob.glob(train_glob)
if len(names) == 0:
break
assert len(names) == 1, 'invalid naming schema in dir'
classifier.sentiment_train(names[0], lt, nb)
test_glob = directory + str(run) + '_test*'
names = glob.glob(test_glob)
assert len(names) == 1, 'invalid naming schema in dir'
result = classifier.sentiment_test(names[0])
print 'Run {0}'.format(run)
print 'Errors: '
for elem in result[3]:
print elem
print '\n\n\n'
f.write('<tr><td>{0}</td>'.format(run))
f.write('<td>{0}</td>'.format(result[0] * 100))
f.write('<td>{0}</td>'.format(result[1] * 100))
f.write('<td>{0}</td></tr>'.format(result[2] * 100))
fp.append(result[0] * 100)
fn.append(result[1] * 100)
acc.append(result[2] * 100)
run += 1
f.write('<tr><td>AVG</td>')
f.write('<td>{0}</td>'.format(sum(fp)/len(fp)))
f.write('<td>{0}</td>'.format(sum(fn)/len(fn)))
f.write('<td>{0}</td></tr>'.format(sum(acc)/len(acc)))
f.write('</tbody></table></html>')
f.close()
def classify_and_write_results(posts):
formatter = Format()
classifier = PostClassifier()
output_file = raw_input("Output file (empty defaults to stdout): ")
try:
f = open(output_file, 'w')
except IOError:
print 'Invalid filename entered. Printing dumped content ... '
f = None
for entry in posts:
entry.sentences = formatter.split_sentences(entry.body)
entry.sentences = zip(
entry.sentences, classifier.classify_sentiment(entry.sentences))
entry.label = classifier.classify_topic_unsupervised(entry.body)
output = '---POST---' + '\nauthor id: ' + str(entry.user_int_id) + \
'\nclass: ' + entry.label + '\nsentences and sentiments:\n' + \
'\n'.join(' : '.join(pair) for pair in entry.sentences)
if (f is not None):
f.write(output)
else:
print output
from classify import PostClassifier
train_path = raw_input("Path to training data: ")
lower_threshold = int(raw_input("Lower threshold: "))
negative_bucket = raw_input("Use negative bucket feature? (True/False): ") == "True"
classifier = PostClassifier()
classifier.sentiment_train(train_path, lower_threshold, negative_bucket)
while True:
sentence = raw_input("Sentence to classify: ")
print "Guess: " + classifier.classify_sentiment(sentence)