tests/testthat/test-draw_data_.R

In sprtt: Sequential Probability Ratio Tests Toolbox

# set.seed(3333)
#
# # null effect ------------------------------------------------------------------
# k_groups <- 4
# f <- 0
# sd <- c(1,1,1,1)
# sample_ratio <-  c(1,1,1,1)
# max_n <- 200
# rep_n <- 100
#
# data <- matrix(
#   data = 0,
#   nrow = max_n*sum(sample_ratio),
#   ncol = rep_n*2
# )
# f_estimated <- double(rep_n)
#
# a = 1
# b = a + 1
# for (i in 1:rep_n) {
#   if (i == 1) {
#     sample <- draw_sample_normal(k_groups = k_groups,
#                                   f = f,
#                                   sd = sd,
#                                   sample_ratio = sample_ratio,
#                                   max_n = max_n)
#     data[, a] <- as.numeric(sample$x)
#     data[, b] <- sample$y
#     f_estimated[i] <- effect_sizes(y~x, sample)$cohens_f
#     a = b + 1
#     b = a + 1
#   } else{
#     sample <- draw_sample_normal(k_groups = k_groups,
#                                   f = f,
#                                   sd = sd,
#                                   sample_ratio = sample_ratio,
#                                   max_n = max_n)
#     f_estimated[i] <- effect_sizes(y~x, sample)$cohens_f
#     data[, a] <- sample$y
#     a = a + 1
#   }
# }
#
# mean_f_est <- mean(f_estimated)
# mean_f_est
#
#
# # small effect -----------------------------------------------------------------
#
# k_groups <- 4
# f <- 0.10
# sd <- c(1,1,1,1)
# sample_ratio <-  c(1,1,1,1)
# max_n <- 200
# rep_n <- 100
#
# data <- matrix(
#   data = 0,
#   nrow = max_n*sum(sample_ratio),
#   ncol = rep_n*2
# )
# f_estimated <- double(rep_n)
#
# a = 1
# b = a + 1
# for (i in 1:rep_n) {
#   if (i == 1) {
#     sample <- draw_sample_normal(k_groups = k_groups,
#                                  f = f,
#                                  sd = sd,
#                                  sample_ratio = sample_ratio,
#                                  max_n = max_n)
#     data[, a] <- as.numeric(sample$x)
#     data[, b] <- sample$y
#     f_estimated[i] <- effect_sizes(y~x, sample)$cohens_f
#     a = b + 1
#     b = a + 1
#   } else{
#     sample <- draw_sample_normal(k_groups = k_groups,
#                                  f = f,
#                                  sd = sd,
#                                  sample_ratio = sample_ratio,
#                                  max_n = max_n)
#     f_estimated[i] <- effect_sizes(y~x, sample)$cohens_f
#     data[, a] <- sample$y
#     a = a + 1
#   }
# }
#
# mean_f_est <- mean(f_estimated)
# mean_f_est
#
#
# # large effect -----------------------------------------------------------------
#
# k_groups <- 4
# f <- 0.80
# sd <- c(1,1,1,1)
# sample_ratio <-  c(1,1,1,1)
# max_n <- 200
# rep_n <- 100
#
# data <- matrix(
#   data = 0,
#   nrow = max_n*sum(sample_ratio),
#   ncol = rep_n*2
# )
# f_estimated <- double(rep_n)
#
# a = 1
# b = a + 1
# for (i in 1:rep_n) {
#   if (i == 1) {
#     sample <- draw_sample_normal(k_groups = k_groups,
#                                  f = f,
#                                  sd = sd,
#                                  sample_ratio = sample_ratio,
#                                  max_n = max_n)
#     data[, a] <- as.numeric(sample$x)
#     data[, b] <- sample$y
#     f_estimated[i] <- effect_sizes(y~x, sample)$cohens_f
#     a = b + 1
#     b = a + 1
#   } else{
#     sample <- draw_sample_normal(k_groups = k_groups,
#                                  f = f,
#                                  sd = sd,
#                                  sample_ratio = sample_ratio,
#                                  max_n = max_n)
#     f_estimated[i] <- effect_sizes(y~x, sample)$cohens_f
#     data[, a] <- sample$y
#     a = a + 1
#   }
# }
#
# mean_f_est <- mean(f_estimated)
# mean_f_est
#
#
# # MIXED ------------------------------------------------------------------------
#
# # large effect -----------------------------------------------------------------
# set.seed(333)
# k_groups <- 4
# f <- 0.70
# sd <- c(1,1,1,1)
# # sd <- c(40,40,40,40)
# sample_ratio <-  c(1,1,1,1)
# max_n <- 200
# rep_n <- 100
#
# data <- matrix(
#   data = 0,
#   nrow = max_n*sum(sample_ratio),
#   ncol = rep_n*2
# )
# f_estimated <- double(rep_n)
#
# a = 1
# b = a + 1
# for (i in 1:rep_n) {
#   if (i == 1) {
#     sample <- draw_sample_mixture(k_groups = k_groups,
#                                  f = f,
#                                  max_n = max_n)
#     data[, a] <- as.numeric(sample$x)
#     data[, b] <- sample$y
#     f_estimated[i] <- effect_sizes(y~x, sample)$cohens_f
#     a = b + 1
#     b = a + 1
#   } else{
#     sample <- draw_sample_mixture(k_groups = k_groups,
#                                  f = f,
#                                  max_n = max_n)
#     f_estimated[i] <- effect_sizes(y~x, sample)$cohens_f
#     data[, a] <- sample$y
#     a = a + 1
#   }
# }
#
# mean_f_est <- mean(f_estimated)
# mean_f_est
# sd_f_est