Chapter 15
15The Hard Realities of Biological Data
15.1Sparsity, imbalance, and noise
Planned: Few labels, rare positives, and measurement error as the default condition.
15.2Batch effects and experimental bias
Planned: Why the confounder is often the assay, the lab, or the cohort ancestry.
15.3Heterogeneity of effects
Planned: When one average hides many mechanisms.