Force R not to use exponential notation (e.g. e+10)?

Question

Can I force R to use regular numbers instead of using the e+10-like notation? I have:

1.810032e+09
# and 
4

within the same vector and want to see:

1810032000
# and
4

I am creating output for an old fashioned program and I have to write a text file using cat. That works fine so far but I simply can't use the e+10 notation there.

Sandhog

Related: https://mcmap.net/q/88935/-number-format-writing-1e-5-instead-of-0-00001/134830 – Askari

Answer 1

This is a bit of a grey area. You need to recall that R will always invoke a print method, and these print methods listen to some options. Including 'scipen' -- a penalty for scientific display. From help(options):

‘scipen’: integer. A penalty to be applied when deciding to print numeric values in fixed or exponential notation. Positive values bias towards fixed and negative towards scientific notation: fixed notation will be preferred unless it is more than ‘scipen’ digits wider.

Example:

R> ran2 <- c(1.810032e+09, 4) 
R> options("scipen"=-100, "digits"=4)
R> ran2
[1] 1.81e+09 4.00e+00
R> options("scipen"=100, "digits"=4)
R> ran2
[1] 1810032000          4

That said, I still find it fudgeworthy. The most direct way is to use sprintf() with explicit width e.g. sprintf("%.5f", ran2).

Rodney

Thanks. scipen seems to be the option I was looking for. The spooky penalty explanation made me shy away. But your example explains it nicely. sprintf, huh? are you referring to the troubles I with sprintf a week ago? :) – Sandhog

In rstudio, if you import a dataset and do train_sample_10k = format(train_sample_10k,scientific=FALSE) and reload, it will change scientific notations. – Whereabouts

How do I return things to normal after having done this? – Scarcely

@CSA: options("scipen"=0, "digits"=7) (those are the default values) – Tripterous

You should move the one that achieves the result options("scipen"=100, "digits"=4) to the top of the code, and the one that doesn't below it ... with the appropriate notes. It can be confusing to someone who is looking for a quick solution (and Google shows the first one as the result). – Instinct

Answer 2

It can be achieved by disabling scientific notation in R.

options(scipen = 999)

Placement

Moreover, this can be put in your .Rprofile file so it gets auto-executed by default. – Hollyhock

If you want to do this just within a function then use withr::local_options(list(scipen = 999)) in the function code. – Kiss

Answer 3

My favorite answer:

format(1810032000, scientific = FALSE)
# [1] "1810032000"

This gives what you want without having to muck about in R settings.

Note that it returns a character string rather than a number object

Glume

Hm that's weird, it doesn't work for me. I don't get an error, it just still prints sciencific notation. – Dichotomous

Not sure what could be wrong. I checked in a very old (3.1.0) and new (3.4.3) version of R and it works for me in both. Most likely some other setting somewhere is taking precedence or you found a version specific or edge-case bug in R. Is it possible you are feeding it a string in scientific notation rather than a numeric object? That would explain it. – Glume

Perhaps noteworthy that this creates a character instead of number. – Louralourdes

If the numbers in your vector are varying lengths, make sure to use justified = "none" or else there will be spaces padding them to the same length. – Pervious

@LaurenFitch its actually justify="none" not justified, in either case I think its actually trim=TRUE that prevents the addition of padding whitespaces and not thejustify flag – Goldina

format(1e6, scientific=FALSE) returns "1000000" while as.character(1e6) returns "1e+06", so there is a difference between the two methods. – Eliseo

This should be the accepted answer as it can be run without manipulation of a global option such as scipen. – Zoogloea

Answer 4

Put options(scipen = 999) in your .Rprofile file so it gets auto-executed by default. (Do not rely on doing it manually.)

(This is saying something different to other answers: how?

1. This keeps things sane when you thunk between multiple projects, multiple languages on a daily or monthly basis. Remembering to type in your per-project settings is error-prone and not scalable. You can have a global ~/.Rprofile or per-project .Rprofile. Or both, with the latter overriding the former.
2. Keeping all your config in a project-wide or global .Rprofile auto-executes it. This is useful for e.g. default package loads, data.table configuration, environment etc. Again, that config can run to a page of settings, and there's zero chance you'll remember those and their syntax and type them in

Hollyhock

Why exactly the same answer? https://mcmap.net/q/88323/-force-r-not-to-use-exponential-notation-e-g-e-10 Apart from the Rprofile bit, maybe better edit the GingerJack's answer? – Hettie

@zx8754: it's not exactly the same answer: the crucial point is move this stuff to your .Rprofile. Then you can never forget it. Also, as time goes by your .Rprofile accumulates all your customizations. – Hollyhock

Up to you of course, but the Q is not "how can I not forget to do X" but "how can I do X". – Hettie

@zx8754: I thunk between R and Python/pandas on multiple projects daily. Both have customizations, package paths etc. It really keeps things sane to have one common config file storing them. Across projects. – Hollyhock

Sorry, I am not questioning usefulness of putting stuff in Rprofile. Anyway, do as you wish, to me your answer looks the same as GingerJack's. Maybe, I am wrong. – Hettie

@zx8754: when you're working on multiple projects across multiple languages, the question "how can I do X" merges with "how can I not forget to do X", in a scalable, consistent, automatic way. I just added more explanation. For whoever the drive-by downvoter is. – Hollyhock