🎬 Movie Reviews Sentiment Analysis Project

A comprehensive sentiment analysis application movie reviews using NLP and machine learning techniques. This project includes both the training notebook and an interactive Streamlit web application.

📋 Project Overview

This project analyzes the sentiment of IMDb movie reviews, classifying them as either positive or negative. It implements two machine learning models:

• Logistic Regression (88.47% accuracy)
• Naive Bayes (85.20% accuracy)

🎯 Key Features

📊 Training & Analysis

• Data Preprocessing: Text cleaning, tokenization, lemmatization
• Feature Extraction: TF-IDF vectorization with 5000 features
• Model Training: Logistic Regression and Naive Bayes classifiers
• Performance Evaluation: Accuracy, precision, recall, F1-score, confusion matrices
• Data Visualization: Word frequency analysis, performance comparisons

🌐 Interactive Web Application

• Project Overview: Complete project documentation and insights
• Training Results: Detailed model performance analysis
• Model Inference: Real-time sentiment prediction interface
• Data Analysis: Dataset exploration and visualization

📁 Project Structure

Sentiment Analysis for IMDb Reviews/
├── Sentiment_Analysis_for_IMDb_Reviews.ipynb  # Training notebook
├── app.py                                     # Streamlit application
├── requirements.txt                           # Python dependencies
├── README.md                                 # Project documentation
├── .gitignore                                # Git ignore file
├── saved_models/                             # Trained models
│   ├── lr_model.pkl                         # Logistic Regression model
│   └── nb_model.pkl                         # Naive Bayes model
└── IMDB Dataset.csv                          # Dataset (not in git)

🚀 Quick Start

Prerequisites

• Python 3.8 or higher
• pip package manager

Installation

1. Clone the repository

   git clone <repository-url>
   cd "Sentiment Analysis for IMDb Reviews"

2. Install dependencies

   pip install -r requirements.txt

3. Download the dataset

   • Place IMDB Dataset.csv in the project root directory
   • The dataset contains 50,000 movie reviews with sentiment labels

4. Run the Streamlit application

   streamlit run app.py

5. Open your browser

   • Navigate to http://localhost:8501
   • Explore the different pages using the sidebar navigation

📖 Usage Guide

🏠 Project Overview

• Complete project documentation
• Technical approach explanation
• Dataset information
• Technology stack details

📊 Training Results

• Model performance comparison
• Detailed metrics and visualizations
• Confusion matrices
• Key insights and analysis

🔍 Model Inference

• Enter a movie review in the text area
• Click "Analyze Sentiment" to get predictions
• View results from both models with confidence scores
• Try example reviews for quick testing

📈 Data Analysis

• Dataset statistics and distribution
• Word frequency analysis
• Text preprocessing insights
• Model comparison details

🔧 Technical Details

Data Preprocessing Pipeline

1. Text Cleaning: Convert to lowercase, remove special characters
2. Tokenization: Split text into individual words
3. Stop Word Removal: Remove common words (the, and, is, etc.)
4. Lemmatization: Convert words to base form
5. TF-IDF Vectorization: Convert to numerical features

Model Architecture

• Logistic Regression: Linear classifier with regularization
• Naive Bayes: Probabilistic classifier based on Bayes theorem
• Feature Space: 5000 TF-IDF features
• Evaluation: 80-20 train-test split

Performance Metrics

Model	Accuracy	Precision	Recall	F1-Score
Logistic Regression	88.47%	0.89	0.88	0.88
Naive Bayes	85.20%	0.85	0.85	0.85

🛠️ Technologies Used

• Python: Core programming language
• NLTK: Natural language processing
• Scikit-learn: Machine learning algorithms
• Pandas: Data manipulation and analysis
• Matplotlib/Seaborn: Data visualization
• Plotly: Interactive visualizations
• Streamlit: Web application framework
• Joblib: Model serialization

📊 Dataset Information

• Source: IMDb movie reviews dataset
• Size: 50,000 reviews
• Classes: Positive (1) and Negative (0)
• Balance: 50% positive, 50% negative
• Features: Text reviews with sentiment labels

🎯 Key Insights

Model Performance

• Logistic Regression performs slightly better overall
• Both models handle balanced datasets well
• TF-IDF features effectively capture sentiment patterns
• Text preprocessing significantly improves performance

Word Analysis

• Positive words: good, story, character, great, see
• Negative words: even, bad, terrible, disappointing, boring
• Some words appear in both categories with different contexts

👨‍💻 Author

Created by Abdelmonem Hatem

🙏 Acknowledgments

• IMDb for providing the dataset
• NLTK and Scikit-learn communities
• Streamlit for the amazing web framework
• Open source community for various libraries and tools

---

Note: Make sure to add the IMDB Dataset.csv file to your .gitignore before pushing to GitHub, as it's not included in this repository due to size constraints.