Text Analysis

November 22, 2024

[1]: ls --all # I have to upload on github so that's why puting Dataset in .ignore␣
↪folder

./ .gitignore/
../ .ipynb_checkpoints/
'Amazon Reviews Analysis.ipynb' RoBERTa-architecture.png
.gi/

[3]: #cd .gitignore/

[5]: # ls # juct checking dataset files

[38]: import pandas as pd

import numpy as np

import nltk
nltk.download('punkt_tab')
nltk.download('averaged_perceptron_tagger_eng')
nltk.download('maxent_ne_chunker_tab')
nltk.download('words')
nltk.download('vader_lexicon')

import seaborn as sb
import matplotlib.pyplot as plt

plt.style.use("ggplot")

[nltk_data] Downloading package punkt_tab to /home/dawood/nltk_data…

[nltk_data] Package punkt_tab is already up-to-date!
[nltk_data] Downloading package averaged_perceptron_tagger_eng to
[nltk_data] /home/dawood/nltk_data…
[nltk_data] Package averaged_perceptron_tagger_eng is already up-to-
[nltk_data] date!
[nltk_data] Downloading package maxent_ne_chunker_tab to
[nltk_data] /home/dawood/nltk_data…
[nltk_data] Package maxent_ne_chunker_tab is already up-to-date!
[nltk_data] Downloading package words to /home/dawood/nltk_data…
[nltk_data] Package words is already up-to-date!
[nltk_data] Downloading package vader_lexicon to

1
 [nltk_data] /home/dawood/nltk_data…
[nltk_data] Package vader_lexicon is already up-to-date!

[252]: df = pd.read_csv(".gitignore/Reviews.csv")

[254]: df = df.head(10000)

[256]: df

[256]: Id ProductId UserId ProfileName \

0 1 B001E4KFG0 A3SGXH7AUHU8GW delmartian
1 2 B00813GRG4 A1D87F6ZCVE5NK dll pa
2 3 B000LQOCH0 ABXLMWJIXXAIN Natalia Corres "Natalia Corres"
3 4 B000UA0QIQ A395BORC6FGVXV Karl
4 5 B006K2ZZ7K A1UQRSCLF8GW1T Michael D. Bigham "M. Wassir"
… … … … …
9995 9996 B000P41A28 A3A63RACXR1XIL A. Boodhoo "deaddodo"
9996 9997 B000P41A28 A5VVRGL8JA7R Adam
9997 9998 B000P41A28 A2TGDTJ8YCU6PD geena77
9998 9999 B000P41A28 AUV4GIZZE693O Susan Coe "sueysis"
9999 10000 B000P41A28 A82WIMR4RSVLI Emrose mom

HelpfulnessNumerator HelpfulnessDenominator Score Time \

0 1 1 5 1303862400
1 0 0 1 1346976000
2 1 1 4 1219017600
3 3 3 2 1307923200
4 0 0 5 1350777600
… … … … …
9995 10 15 1 1204502400
9996 2 3 5 1306368000
9997 0 0 5 1347494400
9998 1 2 5 1203638400
9999 0 1 4 1337472000

Summary \
0 Good Quality Dog Food
1 Not as Advertised
2 "Delight" says it all
3 Cough Medicine
4 Great taffy
… …
9995 constipation
9996 Constipation Not A Problem if…
9997 Love this formula!
9998 very convenient
9999 The best weve tried so far

2
 Text
0 I have bought several of the Vitality canned d…
1 Product arrived labeled as Jumbo Salted Peanut…
2 This is a confection that has been around a fe…
3 If you are looking for the secret ingredient i…
4 Great taffy at a great price. There was a wid…
… …
9995 we switched from the advance similac to the or…
9996 Like the bad reviews say, the organic formula …
9997 I wanted to solely breastfeed but was unable t…
9998 i love the fact that i can get this delieved t…
9999 We have a 7 week old… He had gas and constip…

[10000 rows x 10 columns]

[258]: df.tail(10)

[258]: Id ProductId UserId ProfileName \

9990 9991 B000P41A28 A82CL6H9NWSJC Carl Nothnagel
9991 9992 B000P41A28 A181WVPZSOKTVV GRIZZLY
9992 9993 B000P41A28 A3HINZRNCW1SKA Happy Mom
9993 9994 B000P41A28 AV3IMDC3C0F8 Miss K
9994 9995 B000P41A28 A350OL4V8DV5YK Helen Avramenko
9995 9996 B000P41A28 A3A63RACXR1XIL A. Boodhoo "deaddodo"
9996 9997 B000P41A28 A5VVRGL8JA7R Adam
9997 9998 B000P41A28 A2TGDTJ8YCU6PD geena77
9998 9999 B000P41A28 AUV4GIZZE693O Susan Coe "sueysis"
9999 10000 B000P41A28 A82WIMR4RSVLI Emrose mom

HelpfulnessNumerator HelpfulnessDenominator Score Time \

9990 6 8 5 1302652800
9991 12 17 5 1312675200
9992 1 1 5 1326240000
9993 1 1 5 1202256000
9994 3 4 5 1259539200
9995 10 15 1 1204502400
9996 2 3 5 1306368000
9997 0 0 5 1347494400
9998 1 2 5 1203638400
9999 0 1 4 1337472000

Summary \
9990 Baby loves, and so do we
9991 Good product and highest quality - I'll elabor…
9992 Awesome Similac Organic Formula
9993 It's about time!
9994 great product

3
 9995 constipation
9996 Constipation Not A Problem if…
9997 Love this formula!
9998 very convenient
9999 The best weve tried so far

Text
9990 Great product, our daughter loves this and has…
9991 This is a good product and very high-quality. …
9992 I was NEVER going to give my children formula…
9993 As far as I can tell, this is the only organic…
9994 I switched from Similac Advanced to Organic wh…
9995 we switched from the advance similac to the or…
9996 Like the bad reviews say, the organic formula …
9997 I wanted to solely breastfeed but was unable t…
9998 i love the fact that i can get this delieved t…
9999 We have a 7 week old… He had gas and constip…

[260]: df.shape

[260]: (10000, 10)

[262]: df.Text[0] # just cheching text at 0 index in text column

[262]: 'I have bought several of the Vitality canned dog food products and have found
them all to be of good quality. The product looks more like a stew than a
processed meat and it smells better. My Labrador is finicky and she appreciates
this product better than most.'

[264]: df.Score.value_counts().sort_index().plot(kind="bar", title = "Counts of␣

↪Reviews by Rating", figsize = (8,5))

plt.xlabel("Review stars")
plt.ylabel("Reviews count")
plt.show()

4
[266]: example = df.Text[0]

[268]: example

[268]: 'I have bought several of the Vitality canned dog food products and have found
them all to be of good quality. The product looks more like a stew than a
processed meat and it smells better. My Labrador is finicky and she appreciates
this product better than most.'

[270]: tokens = nltk.word_tokenize(example)

[272]: tokens

[272]: ['I',
'have',
'bought',
'several',
'of',
'the',
'Vitality',
'canned',
'dog',
'food',

5
 'products',
'and',
'have',
'found',
'them',
'all',
'to',
'be',
'of',
'good',
'quality',
'.',
'The',
'product',
'looks',
'more',
'like',
'a',
'stew',
'than',
'a',
'processed',
'meat',
'and',
'it',
'smells',
'better',
'.',
'My',
'Labrador',
'is',
'finicky',
'and',
'she',
'appreciates',
'this',
'product',
'better',
'than',
'most',
'.']

[274]: tagged_tokens = nltk.pos_tag(tokens)

[276]: tagged_tokens

6
[276]: [('I', 'PRP'),
('have', 'VBP'),
('bought', 'VBN'),
('several', 'JJ'),
('of', 'IN'),
('the', 'DT'),
('Vitality', 'NNP'),
('canned', 'VBD'),
('dog', 'RP'),
('food', 'NN'),
('products', 'NNS'),
('and', 'CC'),
('have', 'VBP'),
('found', 'VBN'),
('them', 'PRP'),
('all', 'DT'),
('to', 'TO'),
('be', 'VB'),
('of', 'IN'),
('good', 'JJ'),
('quality', 'NN'),
('.', '.'),
('The', 'DT'),
('product', 'NN'),
('looks', 'VBZ'),
('more', 'RBR'),
('like', 'IN'),
('a', 'DT'),
('stew', 'NN'),
('than', 'IN'),
('a', 'DT'),
('processed', 'JJ'),
('meat', 'NN'),
('and', 'CC'),
('it', 'PRP'),
('smells', 'VBZ'),
('better', 'RBR'),
('.', '.'),
('My', 'PRP$'),
('Labrador', 'NNP'),
('is', 'VBZ'),
('finicky', 'JJ'),
('and', 'CC'),
('she', 'PRP'),
('appreciates', 'VBZ'),
('this', 'DT'),
('product', 'NN'),

7
 ('better', 'JJR'),
('than', 'IN'),
('most', 'JJS'),
('.', '.')]

[278]: entities = nltk.chunk.ne_chunk(tagged_tokens)

entities.pprint()

(S
I/PRP
have/VBP
bought/VBN
several/JJ
of/IN
the/DT
(ORGANIZATION Vitality/NNP)
canned/VBD
dog/RP
food/NN
products/NNS
and/CC
have/VBP
found/VBN
them/PRP
all/DT
to/TO
be/VB
of/IN
good/JJ
quality/NN
./.
The/DT
product/NN
looks/VBZ
more/RBR
like/IN
a/DT
stew/NN
than/IN
a/DT
processed/JJ
meat/NN
and/CC
it/PRP
smells/VBZ
better/RBR
./.
My/PRP$

8
 (ORGANIZATION Labrador/NNP)
is/VBZ
finicky/JJ
and/CC
she/PRP
appreciates/VBZ
this/DT
product/NN
better/JJR
than/IN
most/JJS
./.)

[279]: # VERDAR Sentiment Scoring

[282]: from nltk.sentiment import SentimentIntensityAnalyzer

from tqdm.notebook import tqdm

sia = SentimentIntensityAnalyzer() # just creating object

[284]: sia.polarity_scores("Hi, i am very Happy today.") # compound values are btw -1␣
↪to +1. it's aggregation of neg, neu, pos

[284]: {'neg': 0.0, 'neu': 0.5, 'pos': 0.5, 'compound': 0.6115}

[286]: sia.polarity_scores("Hi, i am very sad today.") # compound values are btw -1 to␣
↪+1. it's aggregation of neg, neu, pos

[286]: {'neg': 0.459, 'neu': 0.541, 'pos': 0.0, 'compound': -0.5256}

[288]: sia.polarity_scores(example) # compound values are btw -1 to +1. it's␣

↪aggregation of neg, neu, pos

[288]: {'neg': 0.0, 'neu': 0.695, 'pos': 0.305, 'compound': 0.9441}

[290]: # now run over the whole dataset

results = {}
for index, row in tqdm(df.iterrows(), total = len(df)):
text = row.Text
id_of_review = row.Id

results[id_of_review] = sia.polarity_scores(text)

0%| | 0/10000 [00:00<?, ?it/s]

[374]: # for i in results.items():

# print(i)

9
[294]: new_df = pd.DataFrame(results).T # to flop horizontally

[296]: new_df = new_df.reindex().rename(columns={"index":"Id"})

[298]: new_df = new_df.merge(df, how="left", left_index=True, right_index=True)

[300]: new_df

[300]: neg neu pos compound Id ProductId UserId \

1 0.000 0.695 0.305 0.9441 2.0 B00813GRG4 A1D87F6ZCVE5NK
2 0.138 0.862 0.000 -0.5664 3.0 B000LQOCH0 ABXLMWJIXXAIN
3 0.091 0.754 0.155 0.8265 4.0 B000UA0QIQ A395BORC6FGVXV
4 0.000 1.000 0.000 0.0000 5.0 B006K2ZZ7K A1UQRSCLF8GW1T
5 0.000 0.552 0.448 0.9468 6.0 B006K2ZZ7K ADT0SRK1MGOEU
… … … … … … … …
9996 0.089 0.852 0.059 -0.5267 9997.0 B000P41A28 A5VVRGL8JA7R
9997 0.091 0.747 0.162 0.6808 9998.0 B000P41A28 A2TGDTJ8YCU6PD
9998 0.063 0.811 0.126 0.9305 9999.0 B000P41A28 AUV4GIZZE693O
9999 0.149 0.697 0.154 0.2809 10000.0 B000P41A28 A82WIMR4RSVLI
10000 0.026 0.811 0.164 0.9850 NaN NaN NaN

ProfileName HelpfulnessNumerator \
1 dll pa 0.0
2 Natalia Corres "Natalia Corres" 1.0
3 Karl 3.0
4 Michael D. Bigham "M. Wassir" 0.0
5 Twoapennything 0.0
… … …
9996 Adam 2.0
9997 geena77 0.0
9998 Susan Coe "sueysis" 1.0
9999 Emrose mom 0.0
10000 NaN NaN

HelpfulnessDenominator Score Time \

1 0.0 1.0 1.346976e+09
2 1.0 4.0 1.219018e+09
3 3.0 2.0 1.307923e+09
4 0.0 5.0 1.350778e+09
5 0.0 4.0 1.342051e+09
… … … …
9996 3.0 5.0 1.306368e+09
9997 0.0 5.0 1.347494e+09
9998 2.0 5.0 1.203638e+09
9999 1.0 4.0 1.337472e+09
10000 NaN NaN NaN

10
 Summary \
1 Not as Advertised
2 "Delight" says it all
3 Cough Medicine
4 Great taffy
5 Nice Taffy
… …
9996 Constipation Not A Problem if…
9997 Love this formula!
9998 very convenient
9999 The best weve tried so far
10000 NaN

Text
1 Product arrived labeled as Jumbo Salted Peanut…
2 This is a confection that has been around a fe…
3 If you are looking for the secret ingredient i…
4 Great taffy at a great price. There was a wid…
5 I got a wild hair for taffy and ordered this f…
… …
9996 Like the bad reviews say, the organic formula …
9997 I wanted to solely breastfeed but was unable t…
9998 i love the fact that i can get this delieved t…
9999 We have a 7 week old… He had gas and constip…
10000 NaN

[10000 rows x 14 columns]

[302]: plot=sb.barplot(data= new_df, x = "Score", y = "compound")

plt.title("Compound Score Review")
plt.show()

11
[303]: plot=sb.barplot(data= new_df, x = "Score", y = "pos")
plt.title("Compound Score Review")
plt.show()

12
[305]: plot=sb.barplot(data= new_df, x = "Score", y = "neu")
plt.title("Compound Score Review")
plt.show()

13
[306]: plot=sb.barplot(data= new_df, x = "Score", y = "neg")
plt.title("Compound Score Review")
plt.show()

14
[308]: # Assuming `new_df` is already defined
fig, axs = plt.subplots(2, 2, figsize=(10, 10)) # Create 2x2 grid

# Plot the bar charts

sb.barplot(data=new_df, x="Score", y="neg", ax=axs[0, 0])
axs[0, 0].set_title("Neg Score Review")

sb.barplot(data=new_df, x="Score", y="compound", ax=axs[0, 1])

axs[0, 1].set_title("Compound Score Review")

sb.barplot(data=new_df, x="Score", y="neu", ax=axs[1, 0])

axs[1, 0].set_title("Neu Score Review")

sb.barplot(data=new_df, x="Score", y="pos", ax=axs[1, 1])

axs[1, 1].set_title("Pos Score Review")

# Adjust layout for better spacing

plt.tight_layout()

15
 # Show the plots
plt.show()

[310]: # RoBERTa

[311]: from transformers import AutoTokenizer

from transformers import AutoModelForSequenceClassification
from scipy.special import softmax

[313]: model = "cardiffnlp/twitter-roberta-base-sentiment-latest"

tokenizer = AutoTokenizer.from_pretrained(model)
model = AutoModelForSequenceClassification.from_pretrained(model)

16
 Some weights of the model checkpoint at cardiffnlp/twitter-roberta-base-
sentiment-latest were not used when initializing
RobertaForSequenceClassification: ['roberta.pooler.dense.bias',
'roberta.pooler.dense.weight']
- This IS expected if you are initializing RobertaForSequenceClassification from
the checkpoint of a model trained on another task or with another architecture
(e.g. initializing a BertForSequenceClassification model from a
BertForPreTraining model).
- This IS NOT expected if you are initializing RobertaForSequenceClassification
from the checkpoint of a model that you expect to be exactly identical
(initializing a BertForSequenceClassification model from a
BertForSequenceClassification model).

[314]: print(example)
sia.polarity_scores(example)

I have bought several of the Vitality canned dog food products and have found
them all to be of good quality. The product looks more like a stew than a
processed meat and it smells better. My Labrador is finicky and she appreciates
this product better than most.

[314]: {'neg': 0.0, 'neu': 0.695, 'pos': 0.305, 'compound': 0.9441}

[315]: encoded_text = tokenizer(example, return_tensors="pt")

[319]: encoded_text

[319]: {'input_ids': tensor([[ 0, 100, 33, 2162, 484, 9, 5, 23911,

1571, 24623,
2335, 689, 785, 8, 33, 303, 106, 70, 7, 28,
9, 205, 1318, 4, 20, 1152, 1326, 55, 101, 10,
24571, 87, 10, 12069, 4884, 8, 24, 28078, 357, 4,
1308, 26882, 16, 8746, 14963, 8, 79, 14137, 1626, 42,
1152, 357, 87, 1437, 144, 4, 2]]), 'attention_mask':
tensor([[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1]])}

[320]: output = model(**encoded_text)

[321]: score = output[0][0].detach().numpy()

[322]: score

[322]: array([-2.0115304, -0.711452 , 2.4868224], dtype=float32)

[323]: score = softmax(score)

[324]: score

17
[324]: array([0.01057769, 0.03881574, 0.9506066 ], dtype=float32)

[325]: score_dic = {
"Neg" : score[0],
"Neu" : score[1],
"Pos" : score[2]
}

[326]: score_dic

[326]: {'Neg': 0.010577693, 'Neu': 0.038815744, 'Pos': 0.9506066}

[330]: def roberta_score(example):

encoded_text = tokenizer(example, return_tensors="pt")
output = model(**encoded_text)
score = output[0][0].detach().numpy()
score = softmax(score)

score_dic = {
'roberta_neg' : score[0],
'roberta_neu' : score[1],
'roberta_pos' : score[2]
}

return score_dic

[332]: roberta_score(example)

[332]: {'roberta_neg': 0.010577693,

'roberta_neu': 0.038815744,
'roberta_pos': 0.9506066}

[340]: res = {}
for i, row in tqdm(df.iterrows(), total=len(df)):
try:
text = row['Text']
myid = row['Id']
vader_result = sia.polarity_scores(text)
vader_result_rename = {}
for key, value in vader_result.items():
vader_result_rename[f"vader_{key}"] = value
roberta_result = roberta_score(text)
both = {**vader_result_rename, **roberta_result}
res[myid] = both
except RuntimeError:
print(f'Broke for id {myid}')

18
 0%| | 0/10000 [00:00<?, ?it/s]
Broke for id 83
Broke for id 187
Broke for id 529
Broke for id 540
Broke for id 746
Broke for id 863
Broke for id 1053
Broke for id 1070
Broke for id 1156
Broke for id 1321
Broke for id 1375
Broke for id 1498
Broke for id 1575
Broke for id 1796
Broke for id 1826
Broke for id 2169
Broke for id 2248
Broke for id 2476
Broke for id 2492
Broke for id 2584
Broke for id 2610
Broke for id 2897
Broke for id 2898
Broke for id 2902
Broke for id 2928
Broke for id 2942
Broke for id 2944
Broke for id 2947
Broke for id 2948
Broke for id 3022
Broke for id 3023
Broke for id 3025
Broke for id 3306
Broke for id 3788
Broke for id 3969
Broke for id 4107
Broke for id 4110
Broke for id 4307
Broke for id 4316
Broke for id 4408
Broke for id 4483
Broke for id 4512
Broke for id 4553
Broke for id 4583
Broke for id 5040
Broke for id 5182

19
 Broke for id 5192
Broke for id 5365
Broke for id 5442
Broke for id 5709
Broke for id 5998
Broke for id 6003
Broke for id 6007
Broke for id 6106
Broke for id 6234
Broke for id 6558
Broke for id 6651
Broke for id 6798
Broke for id 6838
Broke for id 6850
Broke for id 6886
Broke for id 6905
Broke for id 6955
Broke for id 7077
Broke for id 7088
Broke for id 7201
Broke for id 7604
Broke for id 7608
Broke for id 7671
Broke for id 7701
Broke for id 7810
Broke for id 7907
Broke for id 8077
Broke for id 8131
Broke for id 8266
Broke for id 8740
Broke for id 8757
Broke for id 8935
Broke for id 9590
Broke for id 9850
Broke for id 9881
Broke for id 9910
Broke for id 9923

[341]: results_df = pd.DataFrame(res).T

results_df = results_df.reset_index().rename(columns={'index': 'Id'})
results_df = results_df.merge(df, how='left')

[343]: results_df.columns

[343]: Index(['Id', 'vader_neg', 'vader_neu', 'vader_pos', 'vader_compound',

'roberta_neg', 'roberta_neu', 'roberta_pos', 'ProductId', 'UserId',
'ProfileName', 'HelpfulnessNumerator', 'HelpfulnessDenominator',

20
 'Score', 'Time', 'Summary', 'Text'],
dtype='object')

[344]: sb.pairplot(data=results_df,
vars=['vader_neg', 'vader_neu', 'vader_pos', 'roberta_neg',␣
↪'roberta_neu', 'roberta_pos'],

hue='Score',
palette='tab10')
plt.show()

[ ]:

[347]: results_df.query('Score == 1') \

.sort_values('roberta_pos', ascending=False)['Text'].values[0]

21
[347]: 'Bisquick GF is easy to use. Pancakes and muffins are very<br />tasty. The
product is quick and easy to use. It makes my day. Gram'

[348]: results_df.query('Score == 1') \

.sort_values('vader_pos', ascending=False)['Text'].values[0]

[348]: 'This flavor is horrible. There are many other flavors much better. Hawaiian
Hazelnut is great! Breakfast in Bed is AWesome!'

[349]: results_df.query('Score == 5') \

.sort_values('roberta_neg', ascending=False)['Text'].values[0]

[349]: 'They have a bad taste,i finish giving them away because my son started to cry
everytime i tried to give these.I prefer Beech nut or Gerber.Yuck!'

[350]: results_df.query('Score == 5') \

.sort_values('vader_neg', ascending=False)['Text'].values[0]

[350]: 'My two cats must not be interested in grass, because it grew but they ignored
it. Had no problems growing it.'

[351]: from transformers import pipeline

sent_pipeline = pipeline("sentiment-analysis")

No model was supplied, defaulted to distilbert/distilbert-base-uncased-

finetuned-sst-2-english and revision 714eb0f
(https://huggingface.co/distilbert/distilbert-base-uncased-finetuned-
sst-2-english).
Using a pipeline without specifying a model name and revision in production is
not recommended.

[352]: sent_pipeline('I love to coding!')

[352]: [{'label': 'POSITIVE', 'score': 0.999782145023346}]

[353]: sent_pipeline('Make sure you have completed your task!')

[353]: [{'label': 'POSITIVE', 'score': 0.9997491240501404}]

[ ]:

[ ]:

22