Stock Price Prediction Using Machine Learning

Final Project — ML & Data Science

A full-stack web application that fetches live stock data and runs 8 ML models to predict 7-day price forecasts with full evaluation metrics and visualizations.

---

Models Included

#	Model	Type	Library
1	Linear Regression	Baseline linear	scikit-learn
2	Ridge Regression	L2-regularized linear	scikit-learn
3	SVR (RBF kernel)	Support Vector Machine	scikit-learn
4	Random Forest	Ensemble (300 trees)	scikit-learn
5	Gradient Boosting	Boosted ensemble	scikit-learn
6	XGBoost	Extreme gradient boost	xgboost
7	LightGBM	Fast gradient boosting	lightgbm
8	LSTM (Sequence)	Sequence neural net	numpy

XGBoost and LightGBM are the top performers for financial time series.

---

Features Engineered (32 features)

• Moving Averages: MA-5, MA-10, MA-20, MA-50
• EMA-12, EMA-26, MACD, MACD Signal, MACD Histogram
• RSI (14-period)
• ROC-5, ROC-10, Momentum-5
• Bollinger Bands: Upper, Lower, Width, %B
• Volatility (10-day rolling std)
• Price Change %, High-Low %, Close-Open %
• Volume MA, Volume Ratio
• Lag features: Close t-1, t-2, t-3, t-5

Evaluation Metrics

• MSE — Mean Squared Error
• RMSE — Root Mean Squared Error
• R² — Coefficient of Determination (fit quality)
• MAE — Mean Absolute Error

Visualizations

1. 60-day historical price + all 8 model forecasts overlaid
2. 7-day forecast bar chart (best model)
3. Full prediction table (all models × 7 days)
4. Model cards with full metric breakdown
5. R² horizontal bar chart (all models)
6. MSE & RMSE grouped bar chart
7. Test accuracy % bar chart

---

Setup & Run Instructions

Prerequisites

• Python 3.9 or higher
• Internet connection (for live Yahoo Finance data)

---

Step 1 — Unzip the project

cd Downloads
unzip stock_ml_predictor.zip
cd stock_ml_predictor

Step 2 — Create a virtual environment

python -m venv venv

Step 3 — Activate the virtual environment

Windows:

venv\Scripts\activate

Mac / Linux:

source venv/bin/activate

Step 4 — Install all dependencies

pip install -r requirements.txt

This installs Flask, yfinance, scikit-learn, XGBoost, LightGBM, pandas, numpy.

Step 5 — Run the app

python app.py

You'll see output like:

🚀  Stock ML Predictor  →  http://localhost:5000

   XGBoost  : ✓
   LightGBM : ✓

Step 6 — Open browser

http://localhost:5000

---

Usage

1. Enter any stock ticker (AAPL, TSLA, NVDA, AMZN, SPY, etc.)
2. Click Analyze → or press Enter
3. Wait ~20–40 seconds for training 8 models on 2 years of live data
4. Explore all predictions, charts, and model evaluations

---

Project Structure

stock_ml_predictor/
├── app.py                   ← Flask backend + all 8 ML models
├── requirements.txt         ← Python dependencies
├── README.md
├── templates/
│   └── index.html           ← Main HTML page
└── static/
    ├── css/
    │   └── style.css        ← Dark theme UI
    └── js/
        └── main.js          ← Charts (Chart.js) + frontend logic

---

Data Source

Live OHLCV data from Yahoo Finance via yfinance — 2 years of daily data.

---

Final Project — Machine Learning & Data Science