Simulate the Wu & Yao two-component design (eq. 4.1-4.2)

Y = 10 - 10x + e (component 1) or Y = -10 + 10x + e (component 2), with pi = (0.5, 0.5), x ~ U(0, 1), and a bimodal error e ~ 0.5 N(-1, 1^2) + 0.5 N(2, 2^2) (median 0; median regression, tau = 0.5).

Usage

sim_mixqr2(n = 300L, pi1 = 0.5)

Arguments

n

Sample size.

pi1

Probability of component 1.

Value

A data frame with columns x, y, z (true component label).