Power and error#

Relation between power and type I II error#

When we developed a new method or algorithm, we want to evaluate its performance and power. Firstly, we need to evaluate biasness and estimation and expected our model is unbiased in type I error (i.e., \(P(H_{1}|H_{0}) = \alpha\))

Understanding Type I error#

Ref: R type I error

Calculate number of significant variables

Here we use all p-values to get Q-Q plot

Apply \(log_{10}\) function to p-values