| 修订版 | 755b11e0bd678429fa40a6870432fea49cfde86f (tree) |
|---|---|
| 时间 | 2025-01-03 03:35:54 |
| 作者 | Lorenzo Isella <lorenzo.isella@gmai...> |
| Commiter | Lorenzo Isella |
I can now choose whether to run the left_join with arrow or duckplyr.
R-codes/duckplyr_test.R (diff)| @@ -1,9 +1,13 @@ | ||
| 1 | 1 | library(tidyverse) |
| 2 | 2 | |
| 3 | +choose_arrow <- 1 ## choose whether to use arrow or duckplyr | |
| 4 | + | |
| 5 | +if (choose_arrow==1){ | |
| 6 | + | |
| 3 | 7 | library(arrow) |
| 4 | 8 | |
| 5 | 9 | |
| 6 | -# Uncomment and run this only once | |
| 10 | +## Uncomment and run this only once | |
| 7 | 11 | ## dd <- tibble(x=1:100000000, y=rep(LETTERS[1:20], 5000000)) |
| 8 | 12 | |
| 9 | 13 |
| @@ -32,27 +36,29 @@ | ||
| 32 | 36 | |
| 33 | 37 | df_out2|>glimpse() |
| 34 | 38 | |
| 35 | - | |
| 36 | -## uncomment to run --this takes a lot of memory on my system | |
| 37 | - | |
| 38 | -## library(duckplyr) | |
| 39 | +} else { | |
| 39 | 40 | |
| 40 | -## duck_exec("set memory_limit='1GB'") | |
| 41 | 41 | |
| 42 | -## df <- duck_csv("test.csv") | |
| 42 | +library(duckplyr) | |
| 43 | 43 | |
| 44 | -## system.time({ | |
| 45 | -## df_stat <- df |> | |
| 46 | -## summarise(total=sum(x), .by = y) | |
| 44 | +duck_exec("set memory_limit='1GB'") | |
| 45 | + | |
| 46 | +df <- duck_csv("test.csv") | |
| 47 | + | |
| 48 | +system.time({ | |
| 49 | +df_stat <- df |> | |
| 50 | + summarise(total=sum(x), .by = y) | |
| 47 | 51 | |
| 48 | 52 | |
| 49 | 53 | |
| 50 | -## df_out <- df |> | |
| 51 | -## left_join(y=df_stat, by=c("y")) |> | |
| 52 | -## collect() | |
| 54 | +df_out <- df |> | |
| 55 | + left_join(y=df_stat, by=c("y")) |> | |
| 56 | + as_tibble() | |
| 53 | 57 | |
| 54 | -## }) | |
| 58 | +}) | |
| 55 | 59 | |
| 60 | +df_out |> glimpse() | |
| 61 | +} | |
| 56 | 62 | |
| 57 | 63 | sessionInfo() |
| 58 | 64 |
Fork