Data Science Notebooks
Machine learning and deep learning projects in Jupyter Notebook showcasing data engineering, data visualisation and data science skills.
Stock Trading with PyTorch and LSTM networks
A deep learning pipeline and model which uses LSTM networks in PyTorch to predict daily stock price direction (up/down) for S&P 500 stocks. The model uses historical price and volume data from the yfinance package, along with engineered technical indicators, and focuses on directional movement to improve stationarity. Performance is determined by classifying price movements accurately using classification tools from sklearn package, and is benchmarked against perfect foresight and random strategies, with detailed discussion on scalability and real-world limitations of using neural networks for trading stocks.
Bank Churn Analysis
A GitHub repository featuring data analysis and predictive modeling for a bank customer churn dataset (from Kaggle) on Jupyter Notebook. The project addresses the business objective of identifying customers likely to churn, enabling targeted retention strategies. The project includes data preprocessing and exploratory data analysis (EDA) performed using pandas, matplotlib, and other Python libraries, with outputs saved for reproducibility. Classification machine learning techniques with sklearn, including model evaluation and hyperparameter tuning, are applied to predict customer churn. All code and outputs are reproducible.