How to save R data object that has been created using assign()?

Question

I'm creating variables based on certain string combinations. Each variable would store some values. In this example, to make it simple, they store a numerical value. However, in actual problem, each would store a tibble.

I need to store each tibble as RData and they have to be created using unique combinations of the string.

The problem is when I use save() on this variable, it couldn't find it so the save would fail.


res <- 12345
sku = 'sku_a'
index = '1'
# create variable based on string combination
# assign variable value with res
assign(paste0(index,'_arima_',sku), res) 

# return the value of the created variable
get(paste0(index,'_arima_',sku))

# save created variable as RData
save(paste0(index,'_arima_',sku), file = paste0(index,'_arima_',sku,'.RData'))
Error in save(paste0(index, "_arima_", sku), file = paste0(index, "_arima_",  : 
  object ‘paste0(index, "_arima_", sku)’ not found

save(get(paste0(index,'_arima_',sku)), file = paste0(index,'_arima_',sku,'.RData'))
Error in save(get(paste0(index, "_arima_", sku)), file = paste0(index,  : 
  object ‘get(paste0(index, "_arima_", sku))’ not found

save(eval(paste0(index,'_arima_',sku)), file = paste0(index,'_arima_',sku,'.RData'))
Error in save(eval(paste0(index, "_arima_", sku)), file = paste0(index,  : 
  object ‘eval(paste0(index, "_arima_", sku))’ not found

You need a get in your save command. Even better, forget assign (for now) and learn to use lists instead of this assign/get coding nightmare. — Roland
– Roland, Commented Jun 25, 2020 at 11:56
Actually I have no choice but use assign here because of some server issues. Basically the server will randomly crash and I would lose all the previous results unless I store each result from each iteration. And then continue back from the last point when it crashed. I used get() in the second option above, not sure if you can see it. save(get(object)) — Afiq Johari
– Afiq Johari, Commented Jun 25, 2020 at 12:00
@Roland a get in the save command doesn't help here. The first argument of save seems to be captured without being evaluated at all. Your second point is of course spot on. — Allan Cameron
– Allan Cameron, Commented Jun 25, 2020 at 12:15
You can store the results of the iterations in a list just the same as you can store them in the global environment. There is really no reason to use assign in R unless you need fine-control regarding the environment objects are assigned to. You do not have such a case. — Roland
– Roland, Commented Jun 25, 2020 at 13:08
@Roland in the actual code, the results are stored in a list. But the problem with the list is that if the server crashes for whatever reason, the list object will be lost. Hence, I've to rerun the whole loop again. It's probably better if I can save the updated list object at every iteration, instead of saving the result of each iteration in a separate object. — Afiq Johari
– Afiq Johari, Commented Jun 26, 2020 at 1:09

Allan Cameron · Accepted Answer · 2020-06-25 12:22:37Z

2

The first argument of save which is the ... operator is captured unevaluated, so it seems you can't put an expression here. One quick way round this if you have magrittr or dplyr loaded is to use the pipe, which will work:

paste0(index,'_arima_',sku) %>% save(file = paste0(index,'_arima_',sku,'.RData'))

Note though that an added complexity is that variable names shouldn't start with a digit in R. In your case, the variable can only be got from the console by '1_arima_sku_a' (i.e. it needs to be quoted)

answered Jun 25, 2020 at 12:22

Allan Cameron

178k7 gold badges70 silver badges118 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Afiq Johari Over a year ago

Thanks @Allan, thanks for that reminder of starting with a digit. Using piping solves the saving issue. But for some reasons, even though the data is effectively saved, the data doesn't seem to be able to be loaded back to R environment. I think this is a separate question, so I've created a new one here stackoverflow.com/questions/62586733/…

MarinaGA · Accepted Answer · 2023-06-26 17:01:04Z

I ended here trying to avoid my common use of eval(parse(text =...)) and unfortunately, I found that is a good solution for this problem. If 'eval' didn't make it on the original post, it is because the whole expression has to be inside eval, not only the name of the object to be saved:

#### Nedeed original code

res <- 12345
sku = 'sku_a'
index = '1'
assign(paste0(index,'_arima_',sku), res) 
get(paste0(index,'_arima_',sku))

#### Solution to the problem

## save using "eval(parse(text=...))"  
eval(parse(text = paste0("save('",index,"_arima_",sku,
                         "', file = '",index,"_arima_",sku,".RData')")))

## checking
rm(list=ls())
load("1_arima_sku_a.RData")
get("1_arima_sku_a")

Looking forward to hearing a solution to this avoiding 'eval'. In my case, I don't use numbers in the name of the object but still, the name changes on each loop iteration.

Collectives™ on Stack Overflow

How to save R data object that has been created using assign()?

2 Answers 2

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related