MakeUofT 2026: Machine Learning for Gift Gratification
Finding the perfect gift is traditionally a subjective, error-prone task relying heavily on guesswork. At the MakeUofT 2026 Hackathon, the challenge was to quantify human emotion and build a system that removes the ambiguity of gift-giving by objectively measuring a recipient's gratification.
We designed a Biometric Emotion Tracker that pairs physiological hardware with software intelligence. The system monitors live electrical impulses in the skin (GSR) alongside computer-vision facial landmark analysis to gauge real-time happiness, feeding this data into a Random Forest model to optimize future gift recommendations.
This 24-hour hackathon build reinforces my methodology of Accountability through data-driven validation. Furthermore, when our initial heart-rate extraction library failed mid-development, we demonstrated high Efficiency by rapidly pivoting our architecture to utilize OpenCV's DeepFace module instead, securing a functional prototype before the deadline.
We utilized a Grover Galvanic Skin Response (GSR) sensor module to detect electrical impulses in the skin of the gift receiver. This physical data was processed in real-time through an Arduino Uno and an ESP8266 micro-controller, translating physiological responses into quantifiable emotional fluctuations on an integrated LCD.
Computer Vision Integration: After encountering compatibility issues with our initial heart-rate extraction library, we rapidly pivoted our architecture. We implemented OpenCV's DeepFace module to evaluate recipient happiness through facial landmark analysis via a laptop camera.
Machine Learning: The core logic was driven by a Random Forest model trained on a Valentine's gift gratification dataset, effectively correlating user inputs with the highest-performing gift categories to optimize recommendations.