Explainable Survival: From Prediction to Trust in Hex

Local modeling, error analysis, and explainability work supporting an interactive ML app built in Hex.

Problem: Just predicting survival on the Titanic dataset doesn't generate explainable results.

Approach: This project reframes the classic Titanic survival prediction problem as a trust, error analysis, and interpretability challenge...

Tradeoffs: This project was not about producing a Titanic leaderboard 'winner'...

Tech: Python, Hex

View GitHub Repository →← Back to Portfolio