Sentiment Analysis on Restaurant Reviews

Overview

This project focuses on analyzing restaurant reviews to determine whether they are positive or negative. Using machine learning techniques, we classify text reviews based on sentiment. The dataset contains 1000 reviews labeled as either liked (1) or not liked (0).

Dataset

The dataset (Restaurant_Reviews.csv) consists of two columns:

• Review: The text of the restaurant review.
• Liked: A binary label indicating positive (1) or negative (0) sentiment.

Libraries Used

• pandas
• matplotlib
• sklearn (for text vectorization and model training)

Steps Involved

1. Data Exploration

• Loaded the dataset using pandas and checked its structure.
• Plotted basic visualizations to understand sentiment distribution.

2. Data Preprocessing

• Extracted the Review column as input (X) and Liked as the target variable (y).
• Split the dataset into training (80%) and testing (20%).
• Applied TF-IDF Vectorization to convert text into numerical form.

3. Model Training

Two machine learning models were used for classification:

• Logistic Regression: A linear model for binary classification.
• Multinomial Naïve Bayes: A probabilistic classifier suitable for text data.

4. Model Evaluation

• Trained both models on the vectorized review data.
• Evaluated model accuracy using the test dataset.
• The final accuracy achieved was 79.5%.

5. Predicting New Reviews

• The trained model can classify new reviews based on sentiment.
• Example prediction:

  new_text_data = ["Food is bad"]  
  predictions = model.predict(tfidf_vectorizer.transform(new_text_data))  
  print(predictions)  # Output: [0] (negative review)