def return_sentiment(term_vec):
sent_v = []
for term in term_vec:
s = sentiment.sentiment(term)
sent_v.append(
s
) #append each dictionary returned by the sentiment function into a list
return sent_v
def get_average_sentiment(datafr):
""" this function gets the average sentiment for each dataframe """
orig_email = datafr.newcontent.dropna()
# getting rid of http addresses and newline newtab symbols
orig_email = orig_email.apply(lambda x : re.sub(r'http\S+', '', x))
orig_email = orig_email.apply(lambda x : re.sub('\n', '', x))
orig_email = orig_email.apply(lambda x : re.sub('\t', '', x))
sep_email=orig_email.str.split(".").to_frame().values.tolist()
# separating emails into sentences
sentences = []
i = 0
for email in sep_email:
email = email[0]
j = 0
while j < len( email ):
if len( email[ j ] ) == 0:
del email[ j ]
else:
j += 1
for sentence in email:
sentences.append(sentence)
i += 1
# Calulate average valence and arousal for each sentence using Dr. Healey's package
# which uses the ANEW and Happiness sentiment dictionaries
sentiments = [sentiment.sentiment(sentence.split(" ")) for sentence in sentences]
clust0valence = 0
clust0arousal = 0
for x in sentiments:
clust0valence += x['valence']
clust0arousal += x['arousal']
#cluster 0 sentiments
avgval0 = clust0valence/len(sentiments)
avgarousal0 = clust0arousal/len(sentiments)
return [avgval0, avgarousal0]
term_list = []
for term in term_vec_NEW[i]:
if (term not in stop_words and term not in stop_words_2):
term_list.append(term)
term_vec_NEW[i] = term_list
# -----------------------------------------------------------------------
## Term Sentiment
temp_sent = []
sentiments = np.zeros((len(term_vec_NEW), 2))
for i in range(0, len(term_vec_NEW)):
temp_sent = []
temp_sent = str(sentiment.sentiment(term_vec_NEW[i]))
temp_sent = temp_sent.replace('{', '')
temp_sent = temp_sent.replace('}', '')
temp_sent = temp_sent.replace("'", '')
temp_sent = temp_sent.split(sep=',')
temp_sent[0] = temp_sent[0].split(sep=':')
temp_sent[1] = temp_sent[1].split(sep=':')
sentiments[i, 0] = temp_sent[0][1]
sentiments[i, 1] = temp_sent[1][1]
print(np.mean(sentiments[:, 0]))
print(np.mean(sentiments[:, 1]))
cluster_sentiment = cluster_sentiment_fn(sentiments, clusters_LSA)
'''
#document # for the given cluster
#document # for cluster 0
os.chdir(
"C:/Users/Savannah Hampton/Documents/Master of Science in Analytics/Text Analytics/ted-talks"
)
from sentiment_module import sentiment
splittedlist = []
for line in newtedtalks['transcript']:
splittedlist.append(line.split())
for line in splittedlist:
sentiment.exist(line)
sentiment.sentiment(line)
for i in range(0, len(splittedlist)):
print(sentiment.sentiment(splittedlist[i]))
for i in range(2200, 2467):
print(sentiment.sentiment(splittedlist[i]))
#highest valence: 8.13: An electrifying acoustic guitar performance
#average valence: 5.32
#highest arousal: 5.32: An electrifying acoustic guitar performance
#average arousal: 4.51
import pandas as pd
sentiment = pd.read_csv('Sentiment Analysis.csv')
print(sentiment)
#lemmatization
from textblob import Word
stem_df['title'] = stem_df['title'].apply(lambda x: ([Word(word).lemmatize() for word in x]))
# Sentiment
#change directory at first
import os
os.chdir('C:/Users/Debosmita/Documents/TextMining')
#import sentiment module
from sentiment_module import sentiment
count = 0
for index, row in stem_df.iterrows():
print(sentiment.sentiment(row['title']))
count += 1
if count > 5000:
break
i = 0
for email in sep_email:
email = email[0]
j = 0
while j < len( email ):
if len( email[ j ] ) == 0:
del email[ j ]
else:
j += 1
for sentence in email:
sentences.append(sentence)
i += 1
# Calulate average valence and arousal for each sentence using Dr. Healey's package
# which uses the ANEW and Happiness sentiment dictionaries
sentiments = [sentiment.sentiment(sentence.split(" ")) for sentence in sentences]
''' visualizing '''
valences = [sent["valence"] for sent in sentiments]
valences = [i for i in valences if i != 0]
arousals = [sent["arousal"] for sent in sentiments]
arousals = [i for i in arousals if i != 0]
sentiments_overall = list(zip(valences,arousals))
# high arousal, high valence --> orange #fdae61: 0
# low arousal, high valence --> green #abdda4: 1
# low arousal, low valence --> blue #2b83ba: 2
# high arousal, low valence --> red #d7191c: 3
color = []
for sent in sentiments_overall: