[Text]S8_08_Word_Cloud

Echodot2_Reviews.csv

2.34MB

 

 

 

데이터 출처:

https://www.kaggle.com/datasets/PromptCloudHQ/amazon-echo-dot-2-reviews-dataset

 

In [3]:

import pandas as pd
import string
import nltk # natural language processing (자연어 처리)
from nltk.corpus import stopwords
import gensim # 자연어 처리 중 토큰화
from gensim.utils import simple_preprocess

import matplotlib.pyplot as plt
import seaborn as sns

 

In [4]:

echo_df =  pd.read_csv('Echodot2_Reviews.csv', encoding='utf-8')
echo_df.head(2)

 

Out[4]:

	Rating	Review Date	Configuration Text	Review Text	Review Color	Title	User Verified	Review Useful Count	Declaration Text	Pageurl
0	3	10/3/2017	Echo Dot	Not great speakers	Black	Three Stars	Verified Purchase	NaN	NaN	https://www.amazon.com/All-New-Amazon-Echo-Dot...
1	4	9/26/2017	Echo Dot	Great little gagit	White	Four Stars	Verified Purchase	NaN	NaN	https://www.amazon.com/All-New-Amazon-Echo-Dot...

 

In [5]:

echo_df.dropna(subset=['Review Text'], how='any', axis = 0, inplace=True)

 

In [6]:

# A function to remove punctuations
def remove_punc(text):
  Test_removed_punc = [ char for char in text if char not in string.punctuation ]
  Test_punc_join = ''.join(Test_removed_punc)
  return Test_punc_join

 

In [7]:

# download stopwords (불용어) - 감성분석
# 그 자체로 의미가 없는 것들
nltk.download('stopwords')

 

 

[nltk_data] Downloading package stopwords to /root/nltk_data...
[nltk_data]   Unzipping corpora/stopwords.zip.

Out[7]:

True

 

In [8]:

# Add more stopwords : extend
stop_words = stopwords.words('english')
stop_words.extend(['amazon','Amazon','Echo','echo','device','Dot','dot'])

 

In [9]:

# Remove stopwords and remove short words(less than 2 char)
def preprocess(text):
  result = []
  for token in gensim.utils.simple_preprocess(text):
    if token not in stop_words and len(token) >= 1:
      result.append(token)
  return result

 

In [10]:

echo_df['reviews_unpunc']= echo_df['Review Text'].apply(remove_punc)
echo_df['reviews_unpunc_unstopwords'] = echo_df['reviews_unpunc'].apply(preprocess)

 

In [11]:

# join the words into a string
echo_df['Adj_text'] = echo_df['reviews_unpunc_unstopwords'].apply(lambda x: ' '.join(x))

 

In [12]:

# feeback_column
def feedback_class(row):
  if '4' in str(row) or '5' in str(row):
    return 1
  else:
    return 0

 

In [13]:

echo_df['feedback'] = echo_df['Rating'].apply(feedback_class)
echo_df.head() 

 

Out[13]:

	Rating	Review Date	Configuration Text	Review Text	Review Color	Title	User Verified	Review Useful Count	Declaration Text	Pageurl	reviews_unpunc	reviews_unpunc_unstopwords	Adj_text	feedback
0	3	10/3/2017	Echo Dot	Not great speakers	Black	Three Stars	Verified Purchase	NaN	NaN	https://www.amazon.com/All-New-Amazon-Echo-Dot...	Not great speakers	[great, speakers]	great speakers	0
1	4	9/26/2017	Echo Dot	Great little gagit	White	Four Stars	Verified Purchase	NaN	NaN	https://www.amazon.com/All-New-Amazon-Echo-Dot...	Great little gagit	[great, little, gagit]	great little gagit	1
2	5	9/8/2017	Echo Dot	Awesome 👏🏽	White	Awesome!	Verified Purchase	NaN	NaN	https://www.amazon.com/All-New-Amazon-Echo-Dot...	Awesome 👏🏽	[awesome]	awesome	1
3	5	10/19/2017	Echo Dot	Love my Echo	Black	Five Stars	Verified Purchase	NaN	NaN	https://www.amazon.com/All-New-Amazon-Echo-Dot...	Love my Echo	[love]	love	1
4	5	9/17/2017	Echo Dot	Great device	Black	Five Stars	Verified Purchase	NaN	NaN	https://www.amazon.com/All-New-Amazon-Echo-Dot...	Great device	[great]	great	1

 

 

wordcloud

 

In [14]:

sentences = echo_df['Adj_text'].tolist()
sentences[0:5]

 

Out[14]:

['great speakers', 'great little gagit', 'awesome', 'love', 'great']

 

In [15]:

len(sentences)

 

Out[15]:

6852

 

In [16]:

# 모든 문장을 하나의 리스트에 모음
# print(sentences)

 

In [17]:

# Join all elements in the list into one massive string
# 리스트에 모든 문장을 한 문장으로 뭉침
one_string = ' '.join(sentences)
# one_string

 

In [18]:

# the words which mentioning more frequently
from wordcloud import WordCloud
plt.figure(figsize=(20,10))
plt.imshow(WordCloud().generate(one_string))

 

Out[18]:

<matplotlib.image.AxesImage at 0x7f79a6a1d4c0>

 

 

In [19]:

# positive 
positive_df = echo_df[echo_df['Rating']==5]
positive_df.shape

 

Out[19]:

(4382, 14)

 

In [20]:

sentences = positive_df['Adj_text'].tolist()
one_string = ' '.join(sentences)
plt.figure(figsize=(20,10))
plt.imshow(WordCloud().generate(one_string))

 

Out[20]:

<matplotlib.image.AxesImage at 0x7f79a6dcb520>

 

 

In [21]:

# negative
negative_df = echo_df[echo_df['Rating']==1]
negative_df.shape
sentences = negative_df['Adj_text'].tolist()
one_string = ' '.join(sentences)
plt.figure(figsize=(20,10))
plt.imshow(WordCloud().generate(one_string))

 

Out[21]:

<matplotlib.image.AxesImage at 0x7f79a6e00a00>