tubes.py

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

import spacy
import json,os,uuid
import re
import nltk
from nltk.corpus import stopwords
from wordcloud import WordCloud, STOPWORDS
from sklearn.feature_extraction.text import CountVectorizer
from nltk.tokenize import RegexpTokenizer

from imblearn.over_sampling import SMOTE
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score,classification_report
import xgboost as xgb
from sklearn.ensemble import RandomForestClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.naive_bayes import MultinomialNB

from wordcloud import WordCloud, STOPWORDS
import matplotlib.pyplot as plt
from PIL import Image

import warnings
warnings.filterwarnings('ignore')
nltk.download('stopwords')
nlp = spacy.load("en_core_web_sm", disable=['parser', 'ner'])

data = pd.read_csv('chatgpt_reviews.csv')

data.head()

data.info()

data.describe()

data.describe(include='object')

"""<h3> Analysis of Rating column </h3>"""

data['rating'].value_counts().sort_index()

data['rating'].value_counts(normalize=True).mul(100).round(2).sort_index()

#Plot
palette = "deep"
sns.set_palette(palette)

sns.countplot(data=data, x='rating')

plt.xlabel('Rating')
plt.ylabel('No. of Users')
plt.title('Ratings Distribution')

plt.show()

"""Preprocessing"""

#Find no. of missing values in each column
data.isnull().sum().sort_values(ascending=False)

#Combine Review Time and Review
data['complete_review'] = data['title'] +' .'+data['review']
data = data.drop(['date','review','title'],axis='columns')

data.head()

def preprocess_data(text):
    emoji_pattern = re.compile("["
                               u"\U0001F600-\U0001F64F"
                               u"\U0001F300-\U0001F5FF"
                               u"\U0001F680-\U0001F6FF"
                               u"\U0001F1E0-\U0001F1FF"
                               u"\U00002702-\U000027B0"
                               u"\U000024C2-\U0001F251"
                               "]+", flags=re.UNICODE)
    special_char_removal = re.compile(r'[^a-z\d\s]+', re.IGNORECASE)
    text = text.lower()
    text = emoji_pattern.sub('', text)
    text = special_char_removal.sub('', text)
    return text
data['complete_review'] = data['complete_review'].apply(lambda x: preprocess_data(x))
data['complete_review'].head()

preprocess_data("Hallo, My name")

"""hapus stopwords"""

stop = stopwords.words('english')
data['complete_review'] = data['complete_review'].apply(lambda x: " ".join(x for x in x.split() if x not in stop))

"""Lemmatization"""

def space(comment):
    doc = nlp(comment)
    return " ".join([token.lemma_ for token in doc])
data['complete_review']= data['complete_review'].apply(space)

"""menghapus spesifik kata"""

words_to_remove = ['chatgpt','app','chatgpts','chat','gpt','iphone','ipad','gpt4','phone','number','ai','use','io']
data['complete_review'] = data['complete_review'].apply(lambda x: " ".join(x for x in x.split() if x not in words_to_remove))

data['sentiment'] = data['rating'].apply(lambda rating: 1 if rating > 3 else 0)

data.head(5)

data['sentiment'].value_counts(normalize=True).mul(100).round(2)

"""Data is Imbalanced as about 66% of sentiment is positive, 24% is negative and 9.5% is neutral.

# Reviews Analysis
"""

#Analysis of Review field
stopword = set(stopwords.words('english'))
text = " ".join(review for review in data.complete_review)
wordcloud = WordCloud(stopwords=stopword).generate(text)
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis("off")
plt.show()

#positive negative & neutral sentiment:
positive = data[data['sentiment'] == 1]
negative = data[data['sentiment'] == 0]

#Positive Setiment
stopword = set(stopwords.words('english'))
text = " ".join(review for review in positive.complete_review)
wordcloud = WordCloud(stopwords=stopword).generate(text)
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis("off")
plt.show()

#Negative Setiment
stopword = set(stopwords.words('english'))
text = " ".join(review for review in negative.complete_review)
wordcloud = WordCloud(stopwords=stopword).generate(text)
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis("off")
plt.show()

"""Model

Bag of Word Vectorization
"""

#Pre-Prcoessing and Bag of Word Vectorization using Count Vectorizer
token = RegexpTokenizer(r'[a-zA-Z0-9]+')
cv = CountVectorizer(stop_words='english',ngram_range = (1,1),tokenizer = token.tokenize)
X = cv.fit_transform(data['complete_review'])
y = data['sentiment']

"""Handle Imbalanced Data"""

smote = SMOTE()
X_oversampled, y_oversampled = smote.fit_resample(X, y)

"""Train Test Split"""

X_train, X_test, y_train, y_test = train_test_split(X_oversampled,
                                                  y_oversampled,
                                                  test_size=0.15,
                                                  random_state=17,stratify=y_oversampled)

"""XGBoost"""

dtrain = xgb.DMatrix(X_train, label=y_train)
dtest = xgb.DMatrix(X_test, label=y_test)


params = {
    'objective': 'multi:softmax',
    'num_class': 3,
    'eval_metric': 'merror',
    'eta': 0.4,
    'max_depth': 6,
    'subsample': 0.8,
    'colsample_bytree': 0.8,
    'seed': 42
}

num_rounds = 100
model = xgb.train(params, dtrain, num_rounds)

preds = model.predict(dtest)
pred_labels = [int(pred) for pred in preds]

print(classification_report(pred_labels, y_test))

def predict(kata):
    preprocessed_kata = preprocess_data(kata)
    cv_fit = cv.fit(data['complete_review'])
    X_pred = cv_fit.transform(pd.Series([preprocessed_kata]))
    dmatrix = xgb.DMatrix(X_pred)
    preds = model.predict(dmatrix)
    return preds[0]