Reducing a high-dimensional biomedical problem
The source dataset contains 38 explanatory variables covering potentially relevant biomedical attributes. This analysis focuses on a smaller set identified as useful for studying gallstone status, including vitamin D concentration, haemoglobin level, and body-composition measurements.
The project applies multivariate statistical methods to examine dependence between these variables and the binary response. It explores covariance structure, conditional relationships, and the Schur complement as tools for reasoning about correlated biomedical measurements.
The full analysis and reproducible source are available in the linked repository.