def tfidf_gender(db):
female_text = ""
male_text = ""
nb_text = ""
for i in range(1, NUM_OF_PROFILES+1):
# there is an errant frenchman
if i == 1467:
continue
tiptup = db.getText_byID(i)
if tiptup is not None:
if isFemale(i, db):
female_text += '\r'.join(tiptup)
elif isFemale(i, db) is None:
nb_text += '\r'.join(tiptup)
else:
male_text += '\r'.join(tiptup)
corpus = [female_text, male_text, nb_text]
tuple_lists = tf_idf(corpus)
masks = [gal_mask, guy_mask, none_mask]
for i in range(len(tuple_lists)):
generate_tfidf_cloud(tuple_lists[i], masks[i])
def t_test_gender(db, column_name):
dudes = []
ladies = []
for i in range(1, NUM_OF_PROFILES+1):
is_female = isFemale(i, db)
if is_female:
datum = db.getColumn_byID(i, column_name)
if datum:
ladies.append(datum)
elif is_female == False:
datum = db.getColumn_byID(i, column_name)
if datum:
dudes.append(datum)
degrees, t = t_test_math(ladies, dudes)
print(degrees, t)
