🍴 Sentiment Analysis of Restaurant Reviews

This project performs Sentiment Analysis on restaurant reviews using Natural Language Processing (NLP) and Machine Learning (Naïve Bayes Classifier).
It predicts whether a given review is positive 😊 or negative 😞 based on the text entered by the user.

---

📘 Overview

The project uses a dataset of restaurant reviews (Restaurant_Reviews.tsv) to train a text classification model.
By cleaning, tokenizing, stemming, and converting text into numerical form using a Bag of Words model, it learns to classify reviews accurately.

---

🚀 Features

• 🧹 Text preprocessing (cleaning, tokenization, stopword removal, stemming)
• 🧠 Feature extraction using CountVectorizer (Bag of Words)
• 🤖 Trained Naïve Bayes Classifier for sentiment prediction
• ⚙️ Interactive console input to test custom reviews
• 📊 Evaluation metrics: Accuracy, Precision, and Recall
• 🔎 Hyperparameter tuning using different alpha values

---

🧰 Tech Stack

Category	Libraries / Tools Used
Language	Python
Data Handling	NumPy, Pandas
NLP	NLTK
Machine Learning	scikit-learn
Visualization / Output	Console-based metrics and predictions

---

📁 Dataset

The project uses the Restaurant_Reviews.tsv dataset, which contains two columns:

Column	Description
Review	The text review given by a customer
Liked	The sentiment label (1 = Positive, 0 = Negative)

---

⚙️ Installation & Setup

1. Clone the repository

   git clone https://github.com/yourusername/Sentiment_Analysis_of_Restaurant_Reviews.git
   cd Sentiment_Analysis_of_Restaurant_Reviews

2. Install required libraries

   pip install numpy pandas nltk scikit-learn

3. Download NLTK stopwords

   import nltk
   nltk.download('stopwords')

4. Place the dataset
   Ensure the file Restaurant_Reviews.tsv is in the same directory as your Python script.

5. Run the project

   python Sentiment_Analysis_of_Restaurant_Reviews.py

---

🧠 Model Workflow

1. Data Preprocessing

   • Remove non-alphabetic characters using regex
   • Convert text to lowercase
   • Remove English stopwords
   • Apply stemming using PorterStemmer
   • Create a corpus of cleaned reviews

2. Feature Extraction

   • Convert text into numeric vectors using CountVectorizer(max_features=1500)

3. Model Training

   • Split data into training and testing sets (train_test_split)
   • Train using Multinomial Naive Bayes

4. Evaluation

   • Calculate Accuracy, Precision, and Recall
   • Tune alpha hyperparameter to find the best score

5. Prediction

   • Accept custom user input from console
   • Predict whether the review is Positive or Negative

---

📊 Example Output

-------------------
|     SCORES     |
-------------------

Accuracy Score  : 83.5%
Precision Score : 78.6%
Recall Score    : 80.1%

Accuracy Score for alpha=0.1 is : 83.25%
Accuracy Score for alpha=0.2 is : 83.5%
...
The Best Accuracy is 83.5%

----------------------------------------------------------------------------
|  This Model Predicts Whether Restaurant Review is Positive or Negative   |
----------------------------------------------------------------------------

Enter Your Review About Restaurant : The food was delicious and service was great!
=> This is Positive Review.

---

🧪 Example Use

# Example prediction
Enter Your Review About Restaurant : The place was very dirty and food was cold.
=> This is Negative Review.

---

🧾 Notes

• The model is trained using Multinomial Naive Bayes, which is efficient for text classification tasks.
• Adjusting the alpha parameter helps control model smoothing.
• You can experiment with other classifiers like SVM or Logistic Regression for improved accuracy.

---

👩‍💻 Author

Mrunali Badgujar
📧 mrucoder@gmail.com

🔗 GitHub Profile

---

📜 License

This project is licensed under the MIT License — see the LICENSE file for details.

---

✨ “Let the code decide if your meal was worth it!” ✨