# this script creates the dataset for the regression exercise

library(dplyr)
library(faux)

set.seed(123)
n <- 100
dat <-
  rnorm_multi(n = n,
              r = .6,
              varnames = paste0("x", 1:2)) |> 
  mutate(y = x1 + x2 + rnorm(n))

lm(y ~ x1 + x2, dat) |> summary()

dat <-
  dat |> 
  mutate(
    y = ifelse(runif(n) > .5, NA, y)
  )

lm(y ~ x1 + x2, dat) |> summary()

# write.csv(dat, "exercise_01.csv", row.names = F)