10 / Machine learning · Python methodology, browser implementation
Would this passenger have survived?
Explore a live logistic regression model trained in your browser on the actual Titanic training dataset used in my GitHub project. Change a passenger’s profile to recalculate the survival probability immediately.
Model accuracy & validation details
Passenger details
Edit any field, or select a preset to compare passengers.
Survival estimate
What drives this prediction?
Bars show each feature’s signed contribution to the logistic regression score relative to the training-set mean. These are model associations, not causal explanations.
Explore the methodology and limitations
How the model works
Feature preprocessing
Drops passenger ID, name, ticket and cabin. Missing ages are mean-imputed, missing ports use Southampton, and categorical features are encoded numerically. The browser model uses the same seven input features as the original project, with feature standardization for stable optimization.
Data: the 891-row historical Titanic training CSV in the project repository. Training runs locally in the visitor’s browser; no passenger profile is submitted to a server.