Restaurant Review - Sentiment Analysis 🌮: Project Overview

Created a model that can classify a Restaurant Review as a Positive or a Negative review with (77% Accuracy) to detect polarity within the text.

Pulled over 1000 examples from Kaggle using pandas and opendatasets libraries in python.

Applied Logistic Regression, Support Vector Classifier, Random Forest Classifier, Bernoulli Naive Bayes, and KNeighborsClassifier and optimized using GridSearchCV to find the best model.

Code Used

Python version: Python 3.7.11

Packages: pandas, opendatasets, seaborn, matplotlib, numpy, nltk, wordcloud, collections, imblearn.over_sampling, re, string and textblob

Resources Used

The dataset from Kaggle

Functions for Text Data Cleaning

Data Collection

Used Kaggle to pull the datasets 1000 reviews with 2 columns:

• Review
• Liked

Data Cleaning

After pulling the data, I cleaned up the dataset to reduce noises in the dataset. The changes were made follows:

• Made lowercase the sentences, removed punctuations in the sentences, tokenized words, removed stop words from the sentences and lemmatized them.

Exploratory Data Analysis

Visualized the cleaned data to see the trends.

• Created WordCloud for Reviews.

• Created Donut chart for Review data. It looks like our data is balanced.

• Created 2-Gram Analysis Bar Graphs for Review variables.

• Created a histogram for Polarity Score in Sentences Sentences with negative polarity will be in range of [-1, 0), neutral ones will be 0.0, and positive reviews will have the range of (0, 1).

• Created a histogram for Length of Sentences Based on this histogram, we know that our review has text length between approximately 20-80 characters.

• Created a histogram for Word Counts in Sentences From the figure above, we infer that most of the reviews consist of 1 word to 10 words.

Feature Extraction (Vectorization)

Created text features with Term Frequency - Inverse Document Frequency (TF-IDF), Bag-of-Words, and N-Gram then saved them in different dataframes.

Model Building

Data were split into train (80%) and test (20%) sets.

Model Performance Evalution

I used six models (Decision Tree Classifier, Logistic Regression, Support Vector Classifier, Random Forest Classifier, Bernoulli Bayes, and KNeighborsClassifier) to predict the sentiment and evaluated them by using Cross Validation Accuracy Score with three different vectorized data.

I applied cross_val_score to different model with vectorized data combinations to choose the model with the best accuracy score.

Logistic Regression model with TF-IDF vectorized data performed better than any other models in this project.

Model	Cross Validation Accuracy Score
Decision Tree with Bag of Words data	0.7
Decision Tree with TF-IDF data	0.7025
Decision Tree with N-gram data	0.5800
Logistic Regression with Bag of Words data	0.7762
Logistic Regression with TF-IDF data	0.7938
Logistic Regression with N-gram data	0.5713
SVC with Bag of Words data	0.7775
SVC with TF-IDF data	0.7863
SVC with N-gram data	0.58
Random Forest with Bag of Words data	0.7475
Random Forest with TF-IDF data	0.7613
Random Forest with N-gram data	0.5763
Naive Bayes with Bag of Words data	0.7562
Naive Bayes with TF-IDF data	0.7562
Naive Bayes with N-gram data	0.5725
K-Neighbors with Bag of Words data	0.6788
K-Neighbors with TF-IDF data	0.7263
K-Neighbors with N-gram data	0.5163

Hyperparameter Tuning

We got the best accuracy 79.12% with GridSearchCV and find the optimal hyperparameters.

Best Model

Applied Logistic Regression model with the optimal hyperparameters and got 77% Test Accuracy score.

Confusion Matrix

The Confusion Matrix above shows that our model needs to be improved to categorize reviews better.

Since the accuracy on the training data (79%) is higher than the accuracy on the test data (77%), we can say our model is overfitting and needs to be improved.

Thanks for reading :)