4.6 Save and Restore Objects and Workspace

Now, you know how to export and import data frames (or tibbles) to and from various types of files. In this section, you will learn how to save and restore one or more objects that can be of any types, and even the whole workspace that includes all the named objects.

To get started, let’s first clear our workspace using rm(list = ls()) and create a few objects with different types.

rm(list = ls())
dig_num <- 7:1
ani_char <- c("sheep", "pig", "monkey", "pig", "monkey", NA, "pig")

Recall that we can use ls() to get a vector of strings giving the names of the objects in the current environment.

ls()
#> [1] "ani_char" "dig_num"

4.6.1 Save and Restore Objects using .RData

In R, you can use the function save() to save one or more objects into an .RData file. Note that you want to make sure to change the working directory as needed. Let’s see the following example where we save the object dig_num into a file named “dig_num.RData”.

save(dig_num, file = "data/dig_num.RData")

Before introducing how to restore objects, let’s first remove dig_num from our workspace using the rm() function and check its value.

rm(dig_num)
dig_num
#> Error in eval(expr, envir, enclos): object 'dig_num' not found

You can see an error since dig_num has been removed from the workspace. To restore it, you can use the function load() with the corresponding .RData file in double quotes as its argument.

load("data/dig_num.RData")
dig_num
#> [1] 7 6 5 4 3 2 1

You can verify from the value of dig_num that we have successfully restored the object dig_num from the file “dig_num.RData”.

To save more than one objects into a single file, you just need to enter them as additional arguments in the save() function.

save(dig_num, ani_char, file = "data/dig_num_and_ani_char.RData")

To save everything in the workspace, you can use the function save.image() with the desired file name in double quotes as the argument.

save.image("data/all.RData")

To verify that “all.RData” indeed contains all the named object, let’s try the following.

rm(list = ls())  #remove everything from the workspace.
ls()  #confirm the workspace is empty.
#> character(0)
load("data/all.RData")  #restore from 'all.RData'.
ls()  #check what's in the workspace.
#> [1] "ani_char" "dig_num"

4.6.2 Save and Restore a Single Object using saveRDS() and readRDS()

Before introducing the new method, there is one drawback of load() worth noting: if the imported .RData file contains objects with the same names as in the current workspace, all these objects in the current workspace will be silently overwritten without any warning! Let’s see the following example.

dig_num <- 3.14
dig_num
#> [1] 3.14
load("data/all.RData")
dig_num
#> [1] 7 6 5 4 3 2 1

We can see that the value of dig_num was indeed silently overwritten by the load() function, which could be sometimes dangerous.

To avoid this issue, another pair of functions to save and restore a single object is saveRDS() and readRDS(). The usage of saveRDS() is almost identical to save() except we usually use a file with extension “.rds” to store the object.

saveRDS(dig_num, file = "data/dig_num.rds")

To highlight the different behaviors of readRDS() and load(), let’s change the dig_num again.

dig_num <- 666
dig_num
#> [1] 666

To restore the object in an “rds” file, we use the readRDS() in the following way.

dig_num_new <- readRDS("data/dig_num.rds")
dig_num_new
#> [1] 724
dig_num
#> [1] 666

As it is clear from this example, you need to assign the output of the readRDS() function to a name, which helps to prevent any objects been overwritten silently. In fact, the saveRDS() only saves the value of the object without the object name.

For this reason, you are recommended to use the function pair saveRDS() and readRDS() if you want to save and load a single R object.

While save() and load() may be simpler to use when saving and loading multiple objects, you want to be extremely careful with the overwriting issues we discussed here. In fact, you can also use saveRDS() and readRDS() for multiple objects if you create a list containing all the objects.

saveRDS(list(dig_num = dig_num, ani_char = ani_char), file = "data/multi.rds")

Now, let’s try to change the value of dig_num, and recover its original value using readRDS().

dig_num <- 999
dig_num
#> [1] 999
my_list <- readRDS("data/multi.rds")
my_list$dig_num
#> [1] 7 6 5 4 3 2 1

4.6.3 Exercises

1. Define a vector my_vec <- 1:8 and a list my_list <- list(my_num = 5:10, my_char = letters[1:5]). Save my_vec and my_list to a single .RData file named “my_vec_list.RData”. Then clear the workspace, restore these two objects from the .RData file, and verify their values.
2. Save both my_vec and my_list in Q1 to a single .rds file named “my_vec_list.rds”. Then clear the workspace, restore these two objects from the .rds file, and verify their values.