id <- 1:80000
N = 80000
group <- rep(1:8, each = 10000)
group_selection_probability <- c(0.4,0.3,0.15,0.05,0.04,0.03,0.02,0.01)
group_selection_probability <- rep(group_selection_probability, each = 10000)
weight <- (group_selection_probability/sum(group_selection_probability))*N
data <- data.frame(id,group,weight)
sample <- dplyr::sample_n(data, size = 1000, weight = weight)
table(sample$group)