extras/GroupedRankFilter2b.md

In WinVector/rqdatatable: 'rquery' for 'data.table'

Timings of a Grouped Rank Filter Task (extra timings)

Follow-ons to timing article.

Run on a Mac mini (Late 2014), mac OS High Sierra Version 10.13.6, 2.8 GHz Intel Core i5, 8 GB 1600 MHz DD3 RAM, R version 3.5.0, all packages CRAN current as of 8-24-2018 (date of the run).

base_r <- function(df) {
  rownames(df) <- NULL
  df <- df[order(df$col_a, df$col_b, df$col_c, df$col_x, method = 'radix'), , 
           drop = FALSE]
  rownames(df) <- NULL
  n <- length(df$col_a)
  first <- c(TRUE,
             (df$col_a[-1] != df$col_a[-n]) | 
               (df$col_b[-1] != df$col_b[-n]) | 
               (df$col_c[-1] != df$col_c[-n]))
  df <- df[first, , drop = FALSE]
  rownames(df) <- NULL
  df
}

pow <- 6
rds_name <- "GroupedRankFilter2b_runs.RDS"
if(!file.exists(rds_name)) {
  szs <- expand.grid(a = c(1,2,5), b = 10^{0:pow}) 
  szs <- sort(unique(szs$a * szs$b))
  szs <- szs[szs<=10^pow]
  runs <- lapply(
    rev(szs),
    function(sz) {
      gc()
      d <- mk_data(sz)
      ti <- microbenchmark(
        base_r = {
          base_r(d)
        },
        data.table = { 
          # https://stackoverflow.com/questions/16325641/how-to-extract-the-first-n-rows-per-group
          d %.>% 
            as.data.table(.) %.>% 
            setorder(., col_a, col_b, col_c, col_x) %.>%
            .[, .SD[1], by=list(col_a, col_b, col_c)] 
        },
        dplyr = {
          d %>% 
            group_by(col_a, col_b, col_c) %>% 
            arrange(col_x) %>% 
            filter(row_number() == 1) %>%
            ungroup() %>%
            arrange(col_a, col_b, col_c, col_x)
        },
        dplyr_b = {
          d %>% 
            arrange(col_x) %>% 
            group_by(col_a, col_b, col_c) %>% 
            mutate(rn = row_number()) %>%
            ungroup() %>%
            filter(rn == 1) %>%
            select(col_a, col_b, col_c, col_x) %>%
            arrange(col_a, col_b, col_c, col_x)
        },
        dplyr.a = { d %>%
            arrange(col_a, col_b, col_c, col_x) %>%
            group_by(col_a, col_b, col_c) %>%
            slice(1) %>%
            ungroup()
        },

        dplyr.b = {
          d %>%
            arrange(col_a, col_b, col_c, col_x) %>% 
            group_by(col_a, col_b, col_c) %>% 
            mutate(rn = row_number()) %>%
            ungroup() %>%
            filter(rn == 1) %>%
            select(-rn)
        },

        dplyr.c = {
          d %>%
            group_by(col_a, col_b, col_c) %>% 
            summarise(col_x = min(col_x)) %>%
            ungroup() %>%
            arrange(col_a, col_b, col_c, col_x)
        },
        times = 3L,
        check = my_check)
      ti <- as.data.frame(ti)
      ti$rows <- sz
      ti
    })
  saveRDS(runs, rds_name)
} else {
  runs <- readRDS(rds_name)
}

| rows| base_r| data.table| dplyr| dplyr.a| dplyr.b| dplyr.c| dplyr_b| |------:|----------:|-----------:|-----------:|----------:|----------:|----------:|----------:| | 1e+00| 0.0002470| 0.0010742| 0.0029943| 0.0031809| 0.0040186| 0.0022563| 0.0044148| | 2e+00| 0.0003313| 0.0014231| 0.0038599| 0.0031384| 0.0052035| 0.0027281| 0.0056127| | 5e+00| 0.0004026| 0.0014639| 0.0048797| 0.0040526| 0.0059370| 0.0036026| 0.0060995| | 1e+01| 0.0004013| 0.0012798| 0.0041586| 0.0039139| 0.0063424| 0.0032077| 0.0064551| | 2e+01| 0.0005327| 0.0014933| 0.0041455| 0.0039698| 0.0060140| 0.0036182| 0.0065667| | 5e+01| 0.0004635| 0.0014892| 0.0053626| 0.0037429| 0.0062610| 0.0032970| 0.0080441| | 1e+02| 0.0004425| 0.0015018| 0.0054633| 0.0038284| 0.0069292| 0.0027485| 0.0063305| | 2e+02| 0.0004549| 0.0014472| 0.0064043| 0.0037603| 0.0051118| 0.0032940| 0.0065047| | 5e+02| 0.0005361| 0.0015776| 0.0104993| 0.0050768| 0.0059973| 0.0035378| 0.0063656| | 1e+03| 0.0007544| 0.0017001| 0.0176671| 0.0062135| 0.0078708| 0.0046064| 0.0081509| | 2e+03| 0.0009769| 0.0019333| 0.0329028| 0.0078628| 0.0078217| 0.0048941| 0.0113724| | 5e+03| 0.0019663| 0.0024131| 0.0887155| 0.0143565| 0.0127615| 0.0078553| 0.0180241| | 1e+04| 0.0036559| 0.0035723| 0.1394438| 0.0217666| 0.0230061| 0.0148798| 0.0300852| | 2e+04| 0.0074431| 0.0080109| 0.2283989| 0.0502327| 0.0313238| 0.0272954| 0.0536791| | 5e+04| 0.0283719| 0.0126168| 0.5731437| 0.1264591| 0.0980578| 0.0698104| 0.1138654| | 1e+05| 0.0609437| 0.0244684| 1.1894820| 0.2477819| 0.1813457| 0.1461624| 0.2463448| | 2e+05| 0.1138001| 0.0492780| 2.5809338| 0.5841499| 0.4125598| 0.3020178| 0.5090304| | 5e+05| 0.2238009| 0.1433038| 6.9091637| 1.5782126| 1.1714595| 0.9569284| 1.4743319| | 1e+06| 0.6558543| 0.3023460| 14.8725800| 3.7131609| 2.7017660| 1.8995092| 3.4221893|

Full code here.

WinVector/rqdatatable documentation built on Aug. 22, 2023, 3:25 p.m.