2

I have a .CSV file that has two columns one for Tweet and the other for sentiment value formatted like so (but for thousands of tweets):

I like stackoverflow,Positive
Thanks for your answers,Positive
I hate sugar,Negative
I do not like that movie,Negative
stackoverflow is a question and answer site,Neutral
Python is oop high-level programming language,Neutral

I would like to get the output like this:

negfeats = [('I do not like that movie','Negative'),('I hate sugar','Negative')]
posfeats = [('I like stackoverflow','Positive'),('Thanks for your answers','Positive')]
neufeats = [('stackoverflow is a question and answer site','Neutral'),('Python is oop high-level programming language','Neutral')]

I have tried this below to do so but I got some missing chars in tuples. Also, how can I keep x, y, and z as an integer and not a float?

import csv
neg = ['Negative']
pos = ['Positive']
neu = ['Neutral']
neg_counter=0
pos_counter=0
neu_counter=0
negfeats = []
posfeats = []
neufeats = []
with open('ff_tweets.csv', 'Ur') as f:
    for k in f:
        if any(word in k for word in neg):
            negfeats = list(tuple(rec) for rec in csv.reader(f, delimiter=','))
            neg_counter+=1
        elif any(word in k for word in pos):
            posfeats = list(tuple(rec) for rec in csv.reader(f, delimiter=','))
            pos_counter+=1
        else:
            neufeats = list(tuple(rec) for rec in csv.reader(f, delimiter=','))
            neu_counter+=1
x = neg_counter * 3/4
y = pos_counter * 3/4
z = neu_counte * 3/4
print negfeats 
print posfeats 
print neufeats 
print x
print y
print z
Bennett
  • 1,007
  • 4
  • 15
  • 29
Alsphere
  • 513
  • 1
  • 7
  • 22

2 Answers2

0

This should work

import csv

neg = 'Negative'
pos = 'Positive'
neu = 'Neutral'
negfeats = []
posfeats = []
neufeats = []

with open('ff_tweets.csv', 'Ur') as f:
    for r in csv.reader(f):
        if r[1] == neg:
            negfeats.append((r[0], r[1]))
        if r[1] == pos:
            posfeats.append((r[0], r[1]))
        if r[1] == neu:
            neufeats.append((r[0], r[1]))

x = len(negfeats) * float(3)/4
y = len(posfeats) * float(3)/4
z = len(neufeats) * float(3)/4

print negfeats 
print posfeats 
print neufeats 
print x
print y
print z
Ébe Isaac
  • 11,563
  • 17
  • 64
  • 97
  • 1
    You should be careful - in python 2.7 (as in the question) `3/4` returns `0`. – will Nov 13 '16 at 18:45
  • @will True, let's just fix that. – Ébe Isaac Nov 13 '16 at 18:46
  • @ÉbeIsaac this solution theoretically should work but when I run it I got empty lists for negfeats , posfeats and neufeats – Alsphere Nov 13 '16 at 19:01
  • Don't use `is` to compare strings. `r[1] is neg` should be `r[1] == neg`. The checks against `pos` and `neu` should also be changed to use `==`. – cco Nov 13 '16 at 22:37
  • @cco Thank you for the suggestion. I've made the edit, but could you elaborate on why so? – Ébe Isaac Nov 14 '16 at 00:42
  • @Alsphere I've just edited the code. Also notice that neg, pos and neu are not the same as in the question post code. – Ébe Isaac Nov 14 '16 at 00:45
  • 2
    Because `is` tests identity and `==` tests equality. You should (in general) only use `is` with singleton objects (`None` is the best example). http://stackoverflow.com/questions/1504717 has a lot of good explanation and more links. – cco Nov 14 '16 at 02:22
0

Try this, using Pandas. 'Sentiment' is a column in the csv file:

import pandas as pd

df = pd.read_csv('ff_tweets.csv')

pos = tuple(df.loc[df['Sentiment'] == 'Positive'].apply(tuple, axis = 1))
neu = tuple(df.loc[df['Sentiment'] == 'Neutral'].apply(tuple, axis = 1))
neg = tuple(df.loc[df['Sentiment'] == 'Negative'].apply(tuple, axis = 1))

print pos, neg, neu

Output:

(('I like stackoverflow', 'Positive'), ('Thanks for your answers', 'Positive')) (('I hate sugar', 'Negative'), ('I do not like that movie', 'Negative')) (('stackoverflow is a question and answer site', 'Neutral'), ('Python is oop high-level programming language', 'Neutral'))
bumbles
  • 191
  • 1
  • 1
  • 11