In langcog/peekbankr: Accessing the Peekbank Database

library(tidyverse)
library(peekbankr)
library(tictoc)

Test 2x2 (compressed/not and rled/not)

n_sims <- 10

test_timing <- function (compress, rle, dataset_name = "pomper_saffran_2016") {
  compressed_con <- connect_to_peekbank(compress = compress)
  tic()
  get_aoi_timepoints(dataset_name = dataset_name, 
                     connection = compressed_con, 
                     rle = rle)
  t <- toc()
  return(t$toc - t$tic)
}

pomper_timing <- expand_grid(compress = c(TRUE, FALSE), 
                      rle = c(TRUE, FALSE), 
                      nsim = 1:n_sims) %>%
  rowwise() %>%
  mutate(t = test_timing(compress = compress, rle = rle))

ggplot(pomper_timing, aes(x = compress, y = t, col = rle)) + 
  stat_summary(fun.data = "mean_cl_boot") + 
  ylim(0,8) + 
  ylab("Time (s)")

Now try on all data.

all_data_timing <- expand_grid(compress = c(TRUE, FALSE), 
                      rle = c(TRUE, FALSE), 
                      nsim = 1:n_sims) %>%
  rowwise() %>%
  mutate(t = test_timing(compress = compress, rle = rle, dataset_name = NULL))

ggplot(all_data_timing, aes(x = compress, y = t, col = rle)) + 
  stat_summary(fun.data = "mean_cl_boot") + 
  stat_summary(geom= "line") + 
  ylab("Time (s)")

timing <- bind_rows(pomper_timing %>%
                      mutate(dataset = "ps"), 
                    all_data_timing %>%
                      mutate(dataset = "all"))

ggplot(timing, aes(x = compress, y = t, col = rle)) + 
  stat_summary(fun.data = "mean_cl_boot") + 
  stat_summary(geom= "line", aes(group = rle)) + 
  ylab("Time (s)") + 
  scale_y_log10() + 
  facet_wrap(~dataset)