📈 Stock Prediction Tool

An AI-powered stock price prediction app built using Streamlit, LSTM neural networks, and sentiment analysis from news, Twitter, and Google Trends data.

---

🧠 Overview

This tool predicts the next day's stock price using: - Historical price data (via Yahoo Finance) - News sentiment (via FinBERT) - Twitter sentiment (via VADER) - Google Trends data

The model is trained in Google Colab and integrated into the Streamlit app for live inference.

---

🏗️ Project Structure

stock_tool/
├── app.py                   # Streamlit frontend + backend integration
├── backend/
│   ├── stock_data.py        # Fetch stock data from Yahoo Finance
│   ├── sentiment.py         # Fetch sentiment data from news, twitter, google trends
│   ├── features.py          # Feature engineering for ML model
│   ├── model.py             # LSTM model prediction handler
│   ├── train_model.py       # Model training (Google Colab)
├── data/
│   ├── stock_list.csv       # List of stocks with ticker and sector
├── models/                  # Saved trained models and scalers
├── .env                     # API keys and secrets (NOT uploaded to GitHub)
├── requirements.txt         # Dependencies
└── README.md

---

🚀 Features

✅ Predicts next-day stock price using LSTM model
✅ Integrates FinBERT & VADER for sentiment analysis
✅ Uses Google Trends for trend-based features
✅ Sector-wise stock filtering
✅ Provides Buy/Sell/Hold recommendation
✅ Interactive Streamlit dashboard with charts

---

⚙️ Installation

1. Clone the Repository

   git clone https://github.com/Prerna77Arora/Stock_Price_Prediction_Tool.git
   cd Stock_Price_Prediction_Tool

2. Create Virtual Environment

   python -m venv venv
   source venv/bin/activate       # macOS/Linux
   venv\Scripts\activate        # Windows

3. Install Requirements

   pip install -r requirements.txt

4. Add .env File Create a .env file in the project root with your API key:

   NEWS_API_KEY=your_api_key_here

---

🧩 Model Training (in Google Colab)

To retrain models: 1. Open model_training.ipynb or train_model.py in Google Colab
2. Train the LSTM model for multiple stocks
3. The trained .keras models and scalers will be saved in /models

---

🖥️ Run the Streamlit App

streamlit run app.py

Then open the URL shown in your terminal (usually http://localhost:8501).

---

📊 Example Outputs

• Predicted Price: Model's forecast for next trading day\
• Latest Price: Most recent closing price\
• Suggestion: Buy / Sell / Hold decision based on price change threshold

---

🧠 Technologies Used

• Streamlit\
• TensorFlow / Keras\
• Scikit-learn\
• yFinance\
• FinBERT (Transformers)\
• VADER Sentiment Analyzer\
• Google Trends API\
• Plotly for visualization

---

⚠️ Note

• .env file and model weights are not included for security.\
• Some APIs (like NewsAPI) require an API key.

---

👩‍💻 Author

Prerna Arora
B.Tech CSE | Rajiv Gandhi Institute of Petroleum Technology
GitHub: @Prerna77Arora

---