data-raw/DATASET.R

In maoyinan/BayesPC: R package for Bayesian Projection Clustering

## code to prepare `DATASET` dataset goes here

nClusters <- 4
nSub <- 30 * nClusters # subjects
nT <- 40 # time points per subject
dat <- data.frame(
  ID = rep(seq(nSub), each = nT),
  Group = rep(seq(nClusters), each = nSub / nClusters * nT),
  Time = rep(seq(nT), nSub)
) %>%
  mutate(t = (Time - min(Time)) / diff(range((Time))))
dat$Z1 <- 1
for (i in c(1:9)) {
  dat[, paste0("Z", i + 1)] <- cos(pi * i * dat$t)
}
nBasis <- 10

# simulate
set.seed(12345)
epsilon <- rnorm(nSub * nT, 0, 1 / 10)
i1wi <- sample(c(1:3), nSub, replace = T) + 1
i2wi <- sample(c(7:9), nSub, replace = T) + 1

dat$Record <- NA
for (j in 1:nSub) { # filter ID==j
  idx <- which(dat$ID == j)
  gr <- dat$Group[idx[1]] # which cluster determines beta
  beta1 <- ifelse(gr %in% c(1, 2), 1, .1)
  beta2 <- ifelse(gr %in% c(1, 3), 1, .1)
  z1 <- dat[idx, paste0("Z", i1wi[j])]
  z2 <- dat[idx, paste0("Z", i2wi[j])]
  yi <- beta1 * z1 + beta2 * z2 + epsilon[idx]
  dat$Record[idx] <- yi
}
dat$Record <- as.numeric(scale(dat$Record))


DATASET <- dat
use_data(DATASET, overwrite = TRUE)