15.6 Linear regression with interaction

Simulate \(y\) data for a linear regression with one continuous predictor \(x_1\) and one a binary predictor \(x_2\), and an interaction term. This is linear regression with varying intercept and varying slope.

Code

dg$y = 0 + 2*x1 + 3*x2 + 2*x1*x2 + rnorm(n, 0, .1)

g = ggplot(dg, 
           aes(x1, 
               y, 
               group = x2))+
  geom_point(alpha = 0.1)

g %>% 
  pub(xlim = c(0,1))

[1] 20
[1] 80
[1] 20
[1] 120