🎬 IMDB Movie Review Sentiment Analysis

A Streamlit web app that predicts whether a movie review is positive 👍 or negative 👎 using a Simple RNN model trained on the Keras IMDB sentiment dataset.

---

📁 Project Structure

Simple_RNN_imple/
├── app.py                  # Streamlit application
├── SimpleRNN.h5            # Trained Simple RNN sentiment model
├── requirements.txt        # Python dependencies for deployment
├── README.md               # Project documentation
├── SimpleRNN_imple.ipynb   # Model training notebook
├── prediction.ipynb        # Prediction/testing notebook
└── embedding.ipynb         # Word embedding practice notebook

---

✨ Features

• 🔍 Predicts movie review sentiment as positive or negative
• 📊 Shows confidence, positive probability, and negative probability
• 📝 Includes ready-made positive and negative example reviews
• ⚡ Uses cached model loading for smoother Streamlit deployment
• 🧠 Handles unknown words and vocabulary limits consistently with the IMDB dataset

---

⚙️ How It Works

User Input → Lowercase & Tokenize → Map to IMDB Word Index
    → OOV Handling → Pad to 500 tokens → SimpleRNN.h5 → Prediction

1. User enters a movie review in the Streamlit app
2. Text is lowercased and tokenized
3. Tokens are mapped to the Keras IMDB word index
4. Unknown or out-of-vocabulary words are mapped to the OOV token
5. The sequence is padded to 500 tokens
6. SimpleRNN.h5 predicts the positive-class probability
7. Negative probability is calculated as 1 - positive_probability

---

🚀 Run Locally

1. Clone the repository and navigate into the project folder:

git clone <your-repo-url>
cd Simple_RNN_imple

2. Create and activate a virtual environment:

python -m venv venv
source venv/bin/activate      # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Start the Streamlit app:

streamlit run app.py

5. Open in your browser:

http://localhost:8501

---

🧪 Example Sentences

Try these in the app to test the model:

Sentiment	Example
✅ Positive	This movie was fantastic and I loved every minute of it.
❌ Negative	bad awful boring terrible horrible waste worst dull poor disappointing annoying pointless
🔴 Strong Negative	The movie was bad bad bad bad bad bad bad bad bad bad bad bad bad bad bad bad bad bad bad bad.

---

📐 Prediction Logic

The model output is a sigmoid value, representing the probability of the positive class:

positive_probability >= 0.5  →  ✅ Positive
positive_probability  < 0.5  →  ❌ Negative
negative_probability  =  1 - positive_probability

---

🛠️ Tech Stack

Technology	Purpose
🐍 Python	Core language
🔶 TensorFlow / Keras	Model training & inference
🎈 Streamlit	Web application framework
🧠 Simple RNN	Sentiment classification model
📦 Keras IMDB Dataset	Training data

---

📚 References

• Keras IMDB Dataset — https://keras.io/api/datasets/imdb/
• TensorFlow / Keras — https://www.tensorflow.org/
• Streamlit — https://streamlit.io/

---