fst_wakefield.R

library(tidyverse)
library(microbenchmark)
library(wakefield)

simulated_df <- r_data_frame(
  n=10000000,
  id,
  r_series(race,  30, name = "myfactor" ),
  r_series(dob, 30, name = "date"),
  r_series(iq,    30, name = "double")
)


#write 

bench_fwrite <- microbenchmark::microbenchmark(
  data.table::fwrite(simulated_df, "simulated_df.csv"),
  times= 1)

bench_write_rds <- microbenchmark::microbenchmark(
  write_rds(simulated_df, "simulated_df.rds"),
  times = 3)

bench_write_fst <- microbenchmark::microbenchmark(
  fst::write_fst(simulated_df, "simulated_df.fst"),
  times = 3)

bench_write_parquet <- microbenchmark::microbenchmark(
  arrow::write_parquet(simulated_df, "simulated_df.parquet"),
  times =3)

bench_qsave_default <- microbenchmark::microbenchmark(
  qs::qsave(simulated_df, "simulated_df_default.qs", 
            nthreads = parallel::detectCores()),
  times = 3)

bench_qsave_zstd_minus5 <- microbenchmark::microbenchmark(
  qs::qsave(simulated_df, "simulated_df_ztsd_minus5.qs", 
            nthreads =parallel::detectCores(),
            preset = "custom",
            compress_level = -5),
  times = 3)

bench_qsave_zstd_minus30 <- microbenchmark::microbenchmark(
  qs::qsave(simulated_df, "simulated_df_ztsd_minus30.qs", 
            nthreads =parallel::detectCores(),
            preset = "custom",
            compress_level = -30),
  times = 3)

bench_write_feather <- microbenchmark::microbenchmark(
  feather::write_feather(simulated_df, "simulated_df.feather") ,
  times = 3)

benches_write <- bind_rows(
  bench_fwrite,
  bench_write_rds,
  bench_write_fst,
  bench_write_parquet,
  bench_qsave_default,
  bench_qsave_zstd_minus5,
  bench_qsave_zstd_minus30,
  bench_write_feather
)
write_rds(benches_write, "benches_write.rds")
# read
bench_read_csv <-microbenchmark::microbenchmark(
  read_csv("simulated_df.csv"), times = 1)

bench_fread <-  microbenchmark::microbenchmark(
  data.table::fread("simulated_df.csv"), times = 3)


bench_vroom <- microbenchmark::microbenchmark(
  vroom::vroom("simulated_df.csv"),  times=3)

bench_rds <- microbenchmark::microbenchmark(
  readr::read_rds("simulated_df.rds"), times = 5)


bench_fst <-  microbenchmark::microbenchmark(
  fst::read_fst("simulated_df.fst"), times = 10)

bench_parquet <- microbenchmark::microbenchmark(
  arrow::read_parquet("simulated_df.parquet"), times = 10)

bench_qs_default <- microbenchmark::microbenchmark(
  qs::qread("simulated_df_default.qs", nthreads = parallel::detectCores()),  times=10)


bench_qs_ztsd_minus5 <- microbenchmark::microbenchmark(
  qs::qread("simulated_df_ztsd_minus5.qs", nthreads = parallel::detectCores()),  times=10)

bench_qs_ztsd_minus30 <- microbenchmark::microbenchmark(
  qs::qread("simulated_df_ztsd_minus30.qs", nthreads = parallel::detectCores()),  times=10)


bench_feather <-  microbenchmark::microbenchmark(
  feather::read_feather("simulated_df.feather"),  times=10)

benches_read <-dplyr::bind_rows(
  bench_read_csv,
  bench_fread,
  bench_fst,
  bench_rds,
  bench_vroom,
  bench_parquet,
  bench_qs_default,
  bench_qs_ztsd_minus5,
  bench_qs_ztsd_minus30,
  bench_feather)

write_rds(benches_read, "benches_read.rds")