15.7 Poisson regression with interaction

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

Code

dg$y = rpois(n = n, 
             lambda = exp(0 + 0.5*x1 + 1*x2 + 1*x1*x2))

g = ggplot(dg, 
           aes(x = x1, 
               y = y, 
               color = as.factor(x2))) +
  geom_jitter(show.legend = F, 
              alpha  = 0.1, 
              width  = 0,
              height = 0.3) +
  geom_smooth(show.legend = F, 
              linewidth   = 3)

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

[1] 40
[1] 80
[1] 20
[1] 140