By Nitish Adhikari

Email id :[email protected] (mailto:[email protected]), +91-9650740295

Linkedin : https://www.linkedin.com/in/nitish-adhikari-6b2350248 (https://www.linkedin.com/in/nitish-adhikari-6b2350248)

In [1]:

import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt

In [ ]:

comments=pd.read_csv('UScomments.csv',error_bad_lines=False)

In [3]:

comments.head()
Out[3]:

video_id comment_text likes replies

0 XpVt6Z1Gjjo Logan Paul it's yo big day ‼️‼️‼️ 4 0

1 XpVt6Z1Gjjo I've been following you from the start of your... 3 0

2 XpVt6Z1Gjjo Say hi to Kong and maverick for me 3 0

3 XpVt6Z1Gjjo MY FAN . attendance 3 0

4 XpVt6Z1Gjjo trending 😉 3 0

In [4]:

#find out missing values in data

comments.isna().sum()
Out[4]:

video_id 0
comment_text 25
likes 0
replies 0
dtype: int64

In [5]:

##drop missing values as we have very few & update dataframe

comments.dropna(inplace=True)

In [6]:

comments.isna().sum()
Out[6]:

video_id 0
comment_text 0
likes 0
replies 0
dtype: int64

perform Sentiment Analysis

In short , Sentiment analysis is all about analyszing sentiments of Users

In [7]:

# sentiment analysis using TextBlob which is a NLP library built on top of NLTK )..

In [ ]:

!pip install textblob

In [9]:

from textblob import TextBlob

In [10]:

TextBlob('Logan Paul its yo big day ‼️

‼️‼️
').sentiment.polarity
Out[10]:

0.0

In [11]:

df=comments[0:1000]

In [12]:

polarity=[]
for comment in comments['comment_text']:
try:
polarity.append(TextBlob(comment).sentiment.polarity)
except:
polarity.append(0)

In [13]:

print(polarity[0:50])
[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.8, -0.13571428571428573, 0.0, 0.2, -0.023333333333333352, 0.5, 0.0, 0.8,
-0.2916666666666667, 0.0, 0.25, -0.8, 0.0, 0.0, 0.65, 0.0, 0.375, 0.0, 0.0, 0.5, -0.04999999999999999, 0.3444805194
8051944, 0.5, 0.6, 0.0, 0.0, -0.30625, 0.28828125, -0.36458333333333337, 0.5, 0.012499999999999997, 0.1190476190476
1905, 0.16666666666666666, 0.0, -0.4, -0.125, -0.07142857142857142, 0.40727272727272723, 0.0, 0.35, 0.0, -0.0341558
4415584416]

In [14]:

comments.shape
Out[14]:

(691375, 4)

In [15]:

comments.head(3)
Out[15]:

video_id comment_text likes replies

0 XpVt6Z1Gjjo Logan Paul it's yo big day ‼️‼️‼️ 4 0

1 XpVt6Z1Gjjo I've been following you from the start of your... 3 0

2 XpVt6Z1Gjjo Say hi to Kong and maverick for me 3 0

In [16]:

comments['polarity']=polarity

In [17]:

comments.head(12)
Out[17]:

video_id comment_text likes replies polarity

0 XpVt6Z1Gjjo Logan Paul it's yo big day ‼️‼️‼️ 4 0 0.000000

1 XpVt6Z1Gjjo I've been following you from the start of your... 3 0 0.000000

2 XpVt6Z1Gjjo Say hi to Kong and maverick for me 3 0 0.000000

3 XpVt6Z1Gjjo MY FAN . attendance 3 0 0.000000

4 XpVt6Z1Gjjo trending 😉 3 0 0.000000

5 XpVt6Z1Gjjo #1 on trending AYYEEEEE 3 0 0.000000

6 XpVt6Z1Gjjo The end though 😭👍🏻❤️ 4 0 0.000000

7 XpVt6Z1Gjjo #1 trending!!!!!!!!! 3 0 0.000000

8 XpVt6Z1Gjjo Happy one year vlogaversary 3 0 0.800000

9 XpVt6Z1Gjjo You and your shit brother may have single hand... 0 0 -0.135714

10 XpVt6Z1Gjjo There should be a mini Logan Paul too! 0 0 0.000000

11 XpVt6Z1Gjjo Dear Logan, I really wanna get your Merch but ... 0 0 0.200000

Wordcloud Analysis of data

In [1]:

### Lets perform EDA for the Positve sentences

In [19]:

comments_positive=comments[comments['polarity']==1]

In [20]:

comments negative=comments[comments['polarity']==-1]

In [21]:

comments_negative.head(2)
Out[21]:

video_id comment_text likes replies polarity

512 8wNr-NQImFg BEN CARSON IS THE MAN!!!!! THEY HATE HIM CAUSE... 0 0 -1.0

562 8wNr-NQImFg Well… The brain surgeon Ben Carson just proved... 0 0 -1.0

In [ ]:

In [22]:

!pip install wordcloud

In [23]:

from wordcloud import WordCloud , STOPWORDS

In [24]:

comments negative['comment text']

Out[24]:

512 BEN CARSON IS THE MAN!!!!! THEY HATE HIM CAUSE...

562 Well… The brain surgeon Ben Carson just proved...
952 WHY DID YOU MAKE FURRY FORCE?! SO NASTY!!!
1371 WTF BRUH!!!!!!
1391 cheeseus christ thats insane!!!
...
690788 Like Kelly she evil
690865 R U FUCKING KIDDING ME?!?!?!?!
691073 This is horribly offensive please report
691180 Sink holes looks terrifying sinkholes sink you...
691224 Trump talked to the president of US Virgin Isl...
Name: comment_text, Length: 3508, dtype: object

In [25]:

total comments=' '.join(comments negative['comment text'])

In [26]:

total_comments[0:100]
Out[26]:

"BEN CARSON IS THE MAN!!!!! THEY HATE HIM CAUSE HE EXPOSED HITLARY'S RITUAL ABUSE ON CHILDREN!!!!!!! "
In [27]:

wordcloud=WordCloud(stopwords=set(STOPWORDS)).generate(total_comments)
plt.figure(figsize=(15,5))
plt.imshow(wordcloud)
plt.axis('off')
Out[27]:

(-0.5, 399.5, 199.5, -0.5)

In [28]:

# Conclusion-->> Users are emphasizing more on Terrible , worst ,horrible ,boring , disgusting etc

In [ ]:

In [29]:

# perform EDA for the Negative sentences

In [30]:

total comments2=' '.join(comments positive['comment text'])

In [31]:

wordcloud=WordCloud(stopwords=set(STOPWORDS)).generate(total_comments2)
plt.figure(figsize=(15,5))
plt.imshow(wordcloud)
plt.axis('off')
Out[31]:

(-0.5, 399.5, 199.5, -0.5)

In [ ]:

3. Perform Emoji's Analysis

In [32]:

!pip install emoji

Requirement already satisfied: emoji in c:\users\ecotone11\appdata\local\programs\python\python37\lib\site-packages
(2.2.0)

WARNING: There was an error checking the latest version of pip.

In [33]:

import emoji

In [34]:

comments.head(14)
Out[34]:

video_id comment_text likes replies polarity

0 XpVt6Z1Gjjo Logan Paul it's yo big day ‼️‼️‼️ 4 0 0.000000

1 XpVt6Z1Gjjo I've been following you from the start of your... 3 0 0.000000

2 XpVt6Z1Gjjo Say hi to Kong and maverick for me 3 0 0.000000

3 XpVt6Z1Gjjo MY FAN . attendance 3 0 0.000000

4 XpVt6Z1Gjjo trending 😉 3 0 0.000000

5 XpVt6Z1Gjjo #1 on trending AYYEEEEE 3 0 0.000000

6 XpVt6Z1Gjjo The end though 😭👍🏻❤️ 4 0 0.000000

7 XpVt6Z1Gjjo #1 trending!!!!!!!!! 3 0 0.000000

8 XpVt6Z1Gjjo Happy one year vlogaversary 3 0 0.800000

9 XpVt6Z1Gjjo You and your shit brother may have single hand... 0 0 -0.135714

10 XpVt6Z1Gjjo There should be a mini Logan Paul too! 0 0 0.000000

11 XpVt6Z1Gjjo Dear Logan, I really wanna get your Merch but ... 0 0 0.200000

12 XpVt6Z1Gjjo Honestly Evan is so annoying. Like its not fun... 0 0 -0.023333

13 XpVt6Z1Gjjo Casey is still better then logan 0 0 0.500000

In [ ]:

In [35]:

#list of emojis in all comments

emoji_list=[]
for comment in comments['comment_text']:
for char in comment:
if char in emoji.EMOJI_DATA:
emoji_list.append(char)

In [36]:

len(emoji list) #total items in emoji list

Out[36]:

294549

In [37]:

len(pd.Series(emoji list).unique()) #unique items in emoji list

Out[37]:

1098

In [38]:

from collections import Counter

In [39]:

Counter(emoji list) #generate dictionary of count of each emoji

Out[39]:

Counter({'‼': 211,
'😉': 998,
'😭': 8398,
'👍': 5476,
'🏻': 3438,
'❤': 31119,
'😍': 33453,
'💋': 968,
'💙': 2831,
'👈': 126,
'😂': 36987,
'🔥': 8694,
'👻': 268,
'💎': 316,
'😆': 1149,
'😁': 2220,
'😒': 629,
'👏': 5719,

In [40]:

Counter(emoji_list).most_common(10) #10 most common emojis

Out[40]:

[('😂', 36987),
('😍', 33453),
('❤', 31119),
('🔥', 8694),
('😭', 8398),
('👏', 5719),
('😘', 5545),
('👍', 5476),
('💖', 5359),
('💕', 5147)]

In [41]:

#accesing 1st element of list

Counter(emoji_list).most_common(10)[0]
Out[41]:

('😂', 36987)

In [42]:

#accesing 1st item of 1st element of list

Counter(emoji list).most common(10)[0][0]
Out[42]:

'😂'

In [43]:

#Extracting all the emoji from 10 most comman in a list

emojis = [Counter(emoji_list).most_common(10)[i][0] for i in range(10)]
emojis
Out[43]:

['😂', '😍', '❤', '🔥', '😭', '👏', '😘', '👍', '💖', '💕']

In [44]:

#Extracting frequenciesof emoji from 10 most comman in a list

freqs = [Counter(emoji_list).most_common(10)[i][1] for i in range(10)]
freqs
Out[44]:

[36987, 33453, 31119, 8694, 8398, 5719, 5545, 5476, 5359, 5147]

In [45]:

!pip install plotly

Requirement already satisfied: plotly in c:\users\ecotone11\appdata\local\programs\python\python37\lib\site-package
s (5.10.0)
Requirement already satisfied: tenacity>=6.2.0 in c:\users\ecotone11\appdata\local\programs\python\python37\lib\sit
e-packages (from plotly) (8.1.0)

WARNING: There was an error checking the latest version of pip.

In [46]:

import plotly.graph_objects as go
from plotly.offline import iplot

In [47]:

trace = go.Bar(x=emojis,y=freqs)
trace
Out[47]:

Bar({
'x': [😂, 😍, ❤, 🔥, 😭, 👏, 😘, 👍, 💖, 💕],
'y': [36987, 33453, 31119, 8694, 8398, 5719, 5545, 5476, 5359, 5147]
})

In [48]:

iplot([trace])

In [ ]:

Collect Entire Data of Youtube

In [49]:

import os

In [50]:

path = r'E:\Nitish\pd\PJT\Sentiment Analysis\additional_data'

In [51]:

files=os.listdir(path) #list of files in path

files
Out[51]:

['CAvideos.csv',
'CA_category_id.json',
'DEvideos.csv',
'DE_category_id.json',
'FRvideos.csv',
'FR_category_id.json',
'GBvideos.csv',
'GB_category_id.json',
'INvideos.csv',
'IN_category_id.json',
'JPvideos.csv',
'JP_category_id.json',
'KRvideos.csv',
'KR_category_id.json',
'MXvideos.csv',
'MX_category_id.json',
'RUvideos.csv',
'RU_category_id.json',
'USvideos.csv',
'US_category_id.json']

extract list of only .csv files

In [52]:

#extract list of only .csv files

files_csv=[files[i] for i in range(0,len(files),2)]
files csv
Out[52]:

['CAvideos.csv',
'DEvideos.csv',
'FRvideos.csv',
'GBvideos.csv',
'INvideos.csv',
'JPvideos.csv',
'KRvideos.csv',
'MXvideos.csv',
'RUvideos.csv',
'USvideos.csv']

In [53]:

#Extract country name from file name

files_csv[0][0:2]
Out[53]:

'CA'

In [ ]:

#Creating a full dataframe using all the CSV files in the path
full_df = pd.DataFrame()

for file in files_csv:

current_df=pd.read_csv(path+'/'+file,encoding='iso-8859-1',error_bad_lines=False)

current_df['country'] = file[0:2]
full_df = pd.concat([full_df,current_df])
In [55]:

#Created full dataframe

full df.head()
Out[55]:

video_id trending_date title channel_title category_id publish_time tags view

Eminem -
Walk On
2017-11-
0 n1WpP7iowLc 17.14.11 Water EminemVEVO 10 Eminem|"Walk"|"On"|"Water"|"Aftermath/Shady/In... 1715857
10T17:00:03.000Z
(Audio) ft.
Beyoncé

PLUSH -
Bad 2017-11-
1 0dBIkQ4Mz1M 17.14.11 iDubbbzTV 23 plush|"bad unboxing"|"unboxing"|"fan mail"|"id... 101465
Unboxing 13T17:00:00.000Z
Fan Mail

Racist
Superman
| Rudy Rudy 2017-11-
2 5qpjK5DgCt4 17.14.11 23 racist superman|"rudy"|"mancuso"|"king"|"bach"... 319143
Mancuso, Mancuso 12T19:05:24.000Z
King Bach
& Le...

I Dare
You: 2017-11-
3 d380meD0W0M 17.14.11 nigahiga 24 ryan|"higa"|"higatv"|"nigahiga"|"i dare you"|"... 209582
GOING 12T18:01:41.000Z
BALD!?

Ed
Sheeran -
Perfect 2017-11-
4 2Vv-BfVoq4g 17.14.11 Ed Sheeran 10 edsheeran|"ed sheeran"|"acoustic"|"live"|"cove... 3352362
(Official 09T11:04:14.000Z
Music
Video)

In [56]:

full df.shape
Out[56]:

(375942, 17)

Analysing the most liked Category

In [57]:

cat=pd.read_csv('category_file.txt', sep=':')
cat.head()
Out[57]:

Category_id Category_name

1 Film & Animation

2 Autos & Vehicles

10 Music

15 Pets & Animals

17 Sports

In [58]:

cat.reset_index(inplace=True)

In [59]:

cat.columns =['Category_id', 'Category_name']

In [60]:

cat.set index('Category id',inplace=True)

In [61]:

cat
Out[61]:

Category_name

Category_id

1 Film & Animation

2 Autos & Vehicles

10 Music

15 Pets & Animals

17 Sports

18 Short Movies

19 Travel & Events

20 Gaming

21 Videoblogging

22 People & Blogs

23 Comedy

24 Entertainment

25 News & Politics

26 Howto & Style

27 Education

28 Science & Technology

29 Nonprofits & Activism

30 Movies

31 Anime/Animation

32 Action/Adventure

33 Classics

34 Comedy

35 Documentary

36 Drama

37 Family

38 Foreign

39 Horror

40 Sci-Fi/Fantasy

41 Thriller

42 Shorts

43 Shows

44 Trailers
In [62]:

dct=cat.to_dict() #converting dataframe to dictonary

dct
Out[62]:

{'Category_name': {1: ' Film & Animation',

2: ' Autos & Vehicles',
10: ' Music',
15: ' Pets & Animals',
17: ' Sports',
18: ' Short Movies',
19: ' Travel & Events',
20: ' Gaming',
21: ' Videoblogging',
22: ' People & Blogs',
23: ' Comedy',
24: ' Entertainment',
25: ' News & Politics',
26: ' Howto & Style',
27: ' Education',
28: ' Science & Technology',
29: ' Nonprofits & Activism',
30: ' Movies',
31: ' Anime/Animation',
32: ' Action/Adventure',
33: ' Classics',
34: ' Comedy',
35: ' Documentary',
36: ' Drama',
37: ' Family',
38: ' Foreign',
39: ' Horror',
40: ' Sci-Fi/Fantasy',
41: ' Thriller',
42: ' Shorts',
43: ' Shows',
44: ' Trailers '}}

In [63]:

dct['Category name'] #Access category name

Out[63]:

{1: ' Film & Animation',

2: ' Autos & Vehicles',
10: ' Music',
15: ' Pets & Animals',
17: ' Sports',
18: ' Short Movies',
19: ' Travel & Events',
20: ' Gaming',
21: ' Videoblogging',
22: ' People & Blogs',
23: ' Comedy',
24: ' Entertainment',
25: ' News & Politics',
26: ' Howto & Style',
27: ' Education',
28: ' Science & Technology',
29: ' Nonprofits & Activism',
30: ' Movies',
31: ' Anime/Animation',
32: ' Action/Adventure',
33: ' Classics',
34: ' Comedy',
35: ' Documentary',
36: ' Drama',
37: ' Family',
38: ' Foreign',
39: ' Horror',
40: ' Sci-Fi/Fantasy',
41: ' Thriller',
42: ' Shorts',
43: ' Shows',
44: ' Trailers '}

In [64]:

full_df['category_name']=full_df['category_id'].map(dct['Category_name'])
In [65]:

full df.head()
Out[65]:

video_id trending_date title channel_title category_id publish_time tags view

Eminem -
Walk On
2017-11-
0 n1WpP7iowLc 17.14.11 Water EminemVEVO 10 Eminem|"Walk"|"On"|"Water"|"Aftermath/Shady/In... 1715857
10T17:00:03.000Z
(Audio) ft.
Beyoncé

PLUSH -
Bad 2017-11-
1 0dBIkQ4Mz1M 17.14.11 iDubbbzTV 23 plush|"bad unboxing"|"unboxing"|"fan mail"|"id... 101465
Unboxing 13T17:00:00.000Z
Fan Mail

Racist
Superman
| Rudy Rudy 2017-11-
2 5qpjK5DgCt4 17.14.11 23 racist superman|"rudy"|"mancuso"|"king"|"bach"... 319143
Mancuso, Mancuso 12T19:05:24.000Z
King Bach
& Le...

I Dare
You: 2017-11-
3 d380meD0W0M 17.14.11 nigahiga 24 ryan|"higa"|"higatv"|"nigahiga"|"i dare you"|"... 209582
GOING 12T18:01:41.000Z
BALD!?

Ed
Sheeran -
Perfect 2017-11-
4 2Vv-BfVoq4g 17.14.11 Ed Sheeran 10 edsheeran|"ed sheeran"|"acoustic"|"live"|"cove... 3352362
(Official 09T11:04:14.000Z
Music
Video)
Analyse which category has maximum likes

In [66]:

plt.figure(figsize=(12,8))
sns.boxplot(x='category_name',y='likes',data=full_df)
plt.xticks(rotation='vertical')
Out[66]:

(array([ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,

17]),
[Text(0, 0, ' Music'),
Text(1, 0, ' Comedy'),
Text(2, 0, ' Entertainment'),
Text(3, 0, ' News & Politics'),
Text(4, 0, ' People & Blogs'),
Text(5, 0, ' Howto & Style'),
Text(6, 0, ' Film & Animation'),
Text(7, 0, ' Science & Technology'),
Text(8, 0, ' Gaming'),
Text(9, 0, ' Sports'),
Text(10, 0, ' Nonprofits & Activism'),
Text(11, 0, ' Pets & Animals'),
Text(12, 0, ' Travel & Events'),
Text(13, 0, ' Autos & Vehicles'),
Text(14, 0, ' Education'),
Text(15, 0, ' Shows'),
Text(16, 0, ' Movies'),
Text(17, 0, ' Trailers ')])

Analye whether audience is engaged or not

In [67]:

full df.columns
Out[67]:

Index(['video_id', 'trending_date', 'title', 'channel_title', 'category_id',

'publish_time', 'tags', 'views', 'likes', 'dislikes', 'comment_count',
'thumbnail_link', 'comments_disabled', 'ratings_disabled',
'video_error_or_removed', 'description', 'country', 'category_name'],
dtype='object')

In [68]:

#Features that inciates the engagement of audience

full df[['views', 'likes', 'dislikes', 'comment count']].head()
Out[68]:

views likes dislikes comment_count

0 17158579 787425 43420 125882

1 1014651 127794 1688 13030

2 3191434 146035 5339 8181

3 2095828 132239 1989 17518

4 33523622 1634130 21082 85067

In [69]:

#Adding like rate, dislike rate, comment_count_rate to our dataframe

full_df['like_rate']=(full_df['likes']/full_df['views'])*100
full_df['dislike_rate']=(full_df['dislikes']/full_df['views'])*100
full_df['comment_count_rate']=(full_df['comment_count']/full_df['views'])*100

In [70]:

#regression plot for likes vs views

sns.regplot(data=full df, x='views', y='likes')
Out[70]:

<AxesSubplot:xlabel='views', ylabel='likes'>
In [71]:

#Checking correlation between 'views', 'likes', 'dislikes'

sns.heatmap(full df[['views', 'likes', 'dislikes']].corr(),annot=True)
Out[71]:

<AxesSubplot:>

In [ ]:

Which channel has largest number of trending videos

In [72]:

#Channels with highest video_id

full_df.groupby('channel_title')['video_id'].count().sort_values(ascending=False)
Out[72]:

channel_title
The Late Show with Stephen Colbert 984
WWE 804
Late Night with Seth Meyers 773
VikatanTV 763
TheEllenShow 743
...
LIGHTS - 001 jrny 1
bangtanist 1
LIGAMX Femenil 1
LIGA COLOMBIANA OFICIAL 1
Pavel Sidorik TV 1
Name: video_id, Length: 37824, dtype: int64
In [73]:

# Making a dataframe and renaming video_id to total videos

cdf=full_df.groupby('channel_title')['video_id'].count().sort_values(ascending=False).to_frame().reset_index().rename(columns={'v
cdf

Out[73]:

channel_title total_videos

0 The Late Show with Stephen Colbert 984

1 WWE 804

2 Late Night with Seth Meyers 773

3 VikatanTV 763

4 TheEllenShow 743

... ... ...

37819 LIGHTS - 001 jrny 1

37820 bangtanist 1

37821 LIGAMX Femenil 1

37822 LIGA COLOMBIANA OFICIAL 1

37823 Pavel Sidorik TV 1

37824 rows × 2 columns

In [74]:

import plotly.express as px

In [75]:

px.bar(data_frame=cdf[0:20],x='channel_title',y='total_videos')

In [ ]:

Analyse if punctuations in title and tags have any relation with

views,likes,dislikes,comments
In [76]:

import string
In [77]:

string.punctuation
Out[77]:

'!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~'

In [78]:

def punc_count(x):
return len([c for c in x if c in string.punctuation])

In [79]:

#Testing the function

punc_count('The Late Show & with Stephen Colbert')
Out[79]:

In [80]:

#Making a sample dataframe to speed up the comutation time

sample = full df[0:10000]

In [ ]:

#Creating a feature of punctuation count on sample dataframe

sample['count_punc']=sample['title'].apply(punc_count)
sample['count_punc']

In [82]:

sample.head(2) #coun_punc column added to sample dataframe

Out[82]:

video_id trending_date title channel_title category_id publish_time tags views

Eminem -
Walk On
2017-11-
0 n1WpP7iowLc 17.14.11 Water EminemVEVO 10 Eminem|"Walk"|"On"|"Water"|"Aftermath/Shady/In... 17158579
10T17:00:03.000Z
(Audio) ft.
Beyoncé

PLUSH -
Bad 2017-11-
1 0dBIkQ4Mz1M 17.14.11 iDubbbzTV 23 plush|"bad unboxing"|"unboxing"|"fan mail"|"id... 1014651
Unboxing 13T17:00:00.000Z
Fan Mail

2 rows × 22 columns
In [88]:

#creating boxplot for views vs count_puc

plt.figure(figsize=(10,5))
sns.boxplot(x='count punc',y='views',data=sample)
Out[88]:

<AxesSubplot:xlabel='count_punc', ylabel='views'>

In [87]:

sample['count punc'].corr(sample['views'])
Out[87]:

0.0651000978304486

Above fig represents there is 0.06 correlation between puncuation count and views