📊 Stock Sentiment Analyzer using Wikipedia

This project is a pipeline that fetches historical stock data, extracts sentiment from Wikipedia content related to selected companies, and trains a logistic regression model to predict stock movement direction based on sentiment. It also includes a user-friendly Streamlit interface for interaction and visualization.

---

🚀 Features

• ✅ Fetches and updates historical stock data daily using Yahoo Finance
• ✅ Extracts Wikipedia summaries for major tech companies
• ✅ Performs sentiment analysis using VADER on the Wikipedia content
• ✅ Builds a dataset linking sentiment scores to stock price movements
• ✅ Trains a Logistic Regression model to predict upward or downward price movement
• ✅ Saves trained model using joblib
• ✅ Provides a responsive web UI via Streamlit
• ✅ Automatically runs the update pipeline every day at 6 PM
• 🔄 Future scope: pattern-based predictions from stock graphs

---

📂 Folder Structure

GSOC/
├── main.py                 # Main pipeline script (data + training)
├── app.py                  # Streamlit app for UI
├── sentiment_stock_model.joblib  # Trained ML model
├── stock_data.csv          # Historical stock data
├── wiki_sentiment.csv      # Sentiment + direction dataset
├── schedule_task.bat       # Scheduler script for Windows (optional)
├── README.md               # You are here!

---

🛠️ Requirements

Install the required packages using:

pip install -r requirements.txt

Sample requirements.txt:

pandas
yfinance
wikipedia
scikit-learn
vaderSentiment
joblib
streamlit

---

💡 How It Works

1. Update Stock History

Downloads daily stock data for the tickers:

["TSLA", "GOOGL", "AAPL", "AMZN", "MSFT"]

Using Yahoo Finance (yfinance) and stores in stock_data.csv.

---

2. Fetch Wikipedia Sentiment

For each company, fetches up to 3000 characters from its Wikipedia summary and runs VADER sentiment analysis. Example:

• Ticker: TSLA
• Sentiment Score: 0.63

---

3. Train Model

Builds a dataset with:

• Date
• Ticker
• Sentiment Score
• Direction (1 if Close > Open, else 0)

Then trains a Logistic Regression model using scikit-learn.

---

4. Run the Web App

streamlit run app.py

Features:

• Button to run the full pipeline (update → sentiment → train)
• Shows success messages and logs
• Ready to integrate graphs, predictions, and more!

---

🔁 Automating Daily Updates

A scheduled task runs main.py every day at 6 PM.

You can use:

• schedule module for Python automation
• cron (Linux/macOS)
• Task Scheduler (Windows with .bat file)

---

📌 Future Plans

• 📈 Graph-based pattern recognition and predictions
• 📉 Compare sentiment predictions with actual market movement
• 📊 Visual analytics dashboard in Streamlit
• 🌐 Optional API support for integrating external sources (e.g., GDELT, NewsAPI)

---

👨‍💻 Author

Debajeet Mandal
Google Summer of Code 2025 Contributor

---

📃 License

This project is open-sourced under the MIT License.