In MatthiasUckert/Rmatch: A General Package For Any Matching Tasks

library(testthat); library(tidyverse); library(kableExtra); library(scales)

pkgload::load_all(export_all = FALSE)

Check Memory Considerations

make_facke_data <- function(.n, .cols) {
  s_ <- t_ <- tibble::tibble(id = as.character(seq_len(.n)))
  for (i in .cols) {
    s_[[i]] <- purrr::map_chr(paste(seq_len(.n), 1), ~ digest::digest(.x, "xxhash32"))
    t_[[i]] <- purrr::map_chr(paste(seq_len(.n), 2), ~ digest::digest(.x, "xxhash32"))
  }
  return(list(s = s_, t = t_))
}

options(scipen = 999)
cols <- c("name", "city")
n10 <- make_facke_data(10, cols)
n100 <- make_facke_data(100, cols)
n1000 <- make_facke_data(1000, cols)
n10000 <- make_facke_data(10000, cols)

.tmp <- bench::mark(
  n10 = match_data(n10$s, n10$t, cols),
  n100 = match_data(n100$s, n100$t, cols),
  n1000 = match_data(n1000$s, n1000$t, cols),
  n10000 = match_data(n10000$s, n10000$t, cols),
  check = FALSE
) %>% dplyr::select(expression:mem_alloc) 

benchmark <- .tmp %>%
  dplyr::rename(size_mat = expression, mem_calc = mem_alloc) %>%
  dplyr::mutate(
    dplyr::across(c(min, median), as.numeric),
    tmp = as.integer(gsub("n", "", size_mat)),
    dim_mat = paste(comma(tmp), "x", comma(tmp)),
    size_mat = tmp ^ 2,
    mem_calc = as.numeric(mem_calc) / 1e6,
    mem_mat = size_mat * 8 / 1e6,
    `calc/mat` = as.numeric(mem_calc / mem_mat)
  ) %>%
  dplyr::mutate(
    fac_sze = size_mat / lag(size_mat),
    fac_mem = mem_calc / lag(mem_calc),
    fac_time = median / lag(median),
    dplyr::across(!dplyr::matches("_mat$"), ~ round(., 2))
    ) %>%
  dplyr::select(dim_mat, dplyr::everything(), -min, -`itr/sec`, -tmp)

benchmark %>%
  kableExtra::kbl(align = "c", format.args = list(big.mark = ",")) %>%
  kableExtra::kable_paper()

Example Usage

match_cols <- c("name", "iso3", "city", "address")

tab_source <- standardize_data(table_source, match_cols)
tab_target <- standardize_data(table_target, match_cols)

tab_match <- match_data(

)