Project goal is to predict the survival rate of Titanic based on multiple passenger variables .
Use R library Brms to generate stan codes of 3 models:
- Non-hierarchical model with variables: Sex, Age, P-Class
- Non-hierarchical model with all variables
- Hierarchical model with respect to group Sex and Age given P-Class.
Project contains:
-
sensitivity checking with uninformative prior, Normal(0,1e6)and default student's t prior generated by brms.
-
Convergence diagnostics (R-hat, k-value, divergence)
-
Posterior predictive analysis
-
Model comparison (with loo)
Kaggle Predicting Titanic Survival
R labrary brms