The Bias-Variance Tradeoff in High Dimensions – Practice Quiz

Understand why flexible regression can become unstable as predictor dimension approaches sample size. Work with rank, interpolation, bias-variance decomposition, shrinkage and sparse high-dimensional neighborhoods. Prerequisites: linear regression, matrices, expectation and variance. Problems emphasize precise conditions for nonuniqueness, prediction uncertainty, validation and the limits of training fit. Detailed ridge and lasso mechanics are reserved for their later questionsets.

Further reading: ISL reference.