即使设置了 rm(df)，R 循环也会消耗 5GB 的 RAM

Question

I've a loop to make an API call and write a CSV with the data.

After writting it to Local Disk, I've set rm(df). It starts cool, but after 10 or more calls the RAM, and the PC in general starts to slow down very hard.

Windows task manager says RStudio is using 5GB of RAM (total of my PC is 8RM).

Why is this happening if I'm removing the df after it is written to local disk?

library(googleAuthR)
library(googleAnalyticsR)
library(tidyverse)

my_dates <- seq(as.Date("2019-10-25"), as.Date("2019-12-31"), by = 1)


my_fetch <- function(ga_id, d) {



       df <- google_analytics(ga_id,
                   date_range = c(d, d),
                   metrics = c("totalEvents"),
                   dimensions = c("ga:date", "ga:eventCategory", 
                   "ga:eventAction", "ga:eventLabel"),
                   anti_sample = TRUE)



      filename <- paste0("tvgo-web", "-", d, ".csv")
      print(filename)
      write.csv(df, filename, row.names = FALSE)
      rm(df)
      gc()

}



my_fetches_fetches <- mapply(my_fetch, myviewID, my_dates, SIMPLIFY = FALSE)

Each API call as a CSV is 12 MB. It has:

dim(df)
[1] 135410      5

Answer 1

正如@G5W 在评论中发布的那样， rm(df)之后的gc() rm(df)将释放内存。