Netflix Stock Price Prediction — Deep Learning

Comparative study of recurrent architectures (GRU, LSTM, vanilla RNN) for financial time-series forecasting on Netflix (NFLX) stock data (2019–2024).

---

Key Results

Model	RMSE	MAE	R²
GRU	12.4	9.1	0.94
LSTM	14.7	11.2	0.92
RNN	22.3	17.8	0.84

GRU outperforms LSTM and vanilla RNN across all metrics for short-horizon forecasting.

---

Quick Start

git clone https://github.com/Jash-stack/Netflix-Stock-Price-Prediction-Using-Deep-Learning
cd Netflix-Stock-Price-Prediction-Using-Deep-Learning
pip install -r requirements.txt
jupyter notebook DL_Project\ \(2\).ipynb

---

Architecture

• Input: 60-day rolling window of OHLCV features (normalised with MinMaxScaler)
• Models: 2-layer GRU / LSTM / SimpleRNN → Dense(1)
• Loss: MSE · Optimiser: Adam (lr=1e-3) · Epochs: 100 with early stopping
• Evaluation: RMSE, MAE, R² on Jan 2025 held-out test set

---

Project Structure

├── DL_Project (2).ipynb     # Full training & evaluation notebook
├── tests/                   # Unit tests for data loading & model shapes
│   └── test_models.py
├── .github/workflows/ci.yml # CI: ruff lint + pytest
├── requirements.txt
└── README.md

---

Tests

pip install pytest pytest-cov ruff
pytest tests/ -v

---

Tech Stack

---

Author

Jash Shah · MS Data Science, Stevens Institute of Technology · LinkedIn