def get_features(tweet):
features=[]
features.append(len(str(tweet)))
features.append(len(str(tweet).split(" ")))
features.append(len(set(str(tweet))))
features.append(happy(str(tweet)))
features.append(sad(str(tweet)))
features.append(swear_number(str(tweet)))
features.append(pos_emo(str(tweet)))
features.append(neg_emo(str(tweet)))
features=features+feature_getter(str(tweet))
features=features+pos_getter(str(tweet))
features.append(num_stop_words(str(tweet)))
features.append(num_punc(str(tweet)))
return features
def get_features(tweet):
features = []
features.append(timediff(tweet['created_at']))
if ('web' in tweet['source']):
features.append(2)
elif ('mobile' in tweet['source']):
features.append(1)
else:
features.append(0)
if (tweet['coordinates'] == None):
features.append(0)
features.append(0)
else:
a, b = tweet['coordinates']
features.append(a)
features.append(b)
features.append(len(str(tweet['text'].encode('utf-8'))))
features.append(len(str(tweet['text'].encode('utf-8')).split(" ")))
features.append(len(tweet['entities']['hashtags']))
features.append(len(set(str(tweet['text'].encode('utf-8')))))
features.append(len(tweet['entities']['symbols']))
features.append(len(tweet['entities']['urls']))
features.append(happy(str(tweet['text'].encode('utf-8'))))
features.append(sad(str(tweet['text'].encode('utf-8'))))
if (":" in str(tweet['text'].encode('utf-8'))
or ";" in str(tweet['text'].encode('utf-8'))):
features.append(1)
else:
features.append(0)
features.append(swear_number(str(tweet['text'].encode('utf-8'))))
features.append(pos_emo(str(tweet['text'].encode('utf-8'))))
features.append(neg_emo(str(tweet['text'].encode('utf-8'))))
features = features + feature_getter(str(tweet['text'].encode('utf-8')))
features = features + pos_getter(str(tweet['text'].encode('utf-8')))
features.append(tweet['user']['followers_count'])
features.append(tweet['user']['friends_count'])
features.append(timediff(tweet['user']['created_at']))
features.append(tweet['retweet_count'])
features.append(len(tweet['entities']['user_mentions']))
if (tweet['in_reply_to_status_id_str'] != None):
features.append(1)
else:
features.append(0)
if tweet['retweeted']:
features.append(1)
else:
features.append(0)
wot = 0.0
for url in tweet['entities']['urls']:
wot += get_wotscore(str(url))
if (wot == 0):
features.append(wot)
else:
features.append(wot / len(tweet['entities']['urls']))
classifier = raw_input("enter pos or neg or neu:")
if '1' in classifier:
features.append(1)
elif '2' in classifier:
features.append(2)
elif '3' in classifier:
features.append(3)
features.append(num_stop_words(str(tweet['text'].encode('utf-8'))))
features.append(num_punc(str(tweet['text'].encode('utf-8'))))
return features