I'm Thomas Geraci, an applied AI graduate student focused on efficient machine learning, sustainable (green) AI, and real-world predictive modeling.
A supervised machine learning project to predict breast cancer malignancy using diagnostic features from the Wisconsin Breast Cancer Dataset.
Developed as part of the AAI-500 Capstone M.S. Applied Artificial Intelligence, University of San Diego.
Key Highlights:
- 🔎 Data cleaning, preprocessing, and feature selection
- 📈 Logistic Regression & Random Forest modeling
- 📊 Evaluation using Confusion Matrix, ROC Curve, Precision, Recall, F1-Score
- 🛠 Tools: Python, pandas, scikit-learn, seaborn, matplotlib
Team Contributions:
- Denis Mulabegovic: Model building, statistical testing, model documentation
- Thomas Geraci: Data cleaning, exploratory data analysis, visualizations, documentation
- Pros Loung: Model evaluation, classification metrics, final reporting
👉 View Full Project Repository
A machine learning project predicting the severity of forest fires in Portugal’s Montesinho Natural Park using environmental data, regression models, a Keras neural network, and real-time weather API integration.
Developed as part of the AAI-501 Final Project M.S. Applied Artificial Intelligence, University of San Diego.
Key Highlights:
- 🔄 Data cleaning, transformation, and scaling
- 📈 Linear Regression, Random Forest, Lasso Regression, and Keras Neural Network
- 🌦 Integration of 7-day historical weather data from Open Meteo API
- 📊 Evaluation using R² Score and Mean Absolute Error (MAE)
- 🛠 Tools: Python, pandas, scikit-learn, seaborn, matplotlib, TensorFlow/Keras, API integration
Team Contributions:
- Thomas Geraci: Data preprocessing, Linear Regression, residual and prediction-vs-actual plots, Keras and API integration
- Daniel Sims: Random Forest implementation and feature importance analysis
- Arslan Isaac: Lasso Regression and feature selection
- All Members: Keras deep learning model and weather API integration
👉 View Full Project Repository
Feel free to explore my full repositories or connect with me on LinkedIn!