17 messages · Erin Hodgess, Rolf Turner, (Ted Harding) +10 more
Hi R People: I'm teaching a statistical computing class using R starting next week (yay!) and I have an opinion type question, please. I'm old school and use "<-" in an assignment. However, I'm starting to see the "=" in the literature. Which should I use or does it matter, please? Thanks for your input! Sincerely, Erin
Erin Hodgess Associate Professor Department of Computer and Mathematical Sciences University of Houston - Downtown mailto: erinm.hodgess at gmail.com
Hi R People: I'm teaching a statistical computing class using R starting next week (yay!) and I have an opinion type question, please. I'm old school and use "<-" in an assignment. However, I'm starting to see the "=" in the literature. Which should I use or does it matter, please?
It's basically a matter of taste. But people with ***good*** taste
use "<-". :-)
Constructions such as "a = a+3" are toadally illogical whereas "a <- a
+3"
makes perfect sense.
However I'm sore afraid that as is always the case, good taste fails and
bad taste prevails. I.e. "=" for assignment will take over. Personally
I shall resist as long as possible, i.e. until "<-" is removed from the
syntax structure by R Core.
cheers,
Rolf
######################################################################
Attention:\ This e-mail message is privileged and confid...{{dropped:9}}
On 15/01/2010, at 4:45 PM, Erin Hodgess wrote:
Hi R People: I'm teaching a statistical computing class using R starting next week (yay!) and I have an opinion type question, please. I'm old school and use "<-" in an assignment. However, I'm starting to see the "=" in the literature. Which should I use or does it matter, please?
It's basically a matter of taste. But people with ***good*** taste use "<-". :-) Constructions such as "a = a+3" are toadally illogical whereas "a <- a + 3" makes perfect sense. However I'm sore afraid that as is always the case, good taste fails and bad taste prevails. I.e. "=" for assignment will take over.
Yes, water flows down-hill -- unless we build control structures ...
Personally I shall resist as long as possible, i.e. until "<-"
is removed from the syntax structure by R Core.
cheers,
Rolf
There is at least one context where the distinction must be preserved. Example: pnorm(1.5) # [1] 0.9331928 pnorm(x=1.5) # Error in pnorm(x = 1.5) : unused argument(s) (x = 1.5) pnorm(x<-1.5) # [1] 0.9331928 x # [1] 1.5 Ted. -------------------------------------------------------------------- E-Mail: (Ted Harding) <Ted.Harding at manchester.ac.uk> Fax-to-email: +44 (0)870 094 0861 Date: 15-Jan-10 Time: 06:57:16 ------------------------------ XFMail ------------------------------
An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20100115/9b9723b1/attachment.pl>
An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20100115/83a264e5/attachment.pl>
On Fri, Jan 15, 2010 at 6:57 AM, Ted Harding <Ted.Harding at manchester.ac.uk>wrote:
There is at least one context where the distinction must be preserved. Example: pnorm(1.5) # [1] 0.9331928 pnorm(x=1.5) # Error in pnorm(x = 1.5) : unused argument(s) (x = 1.5) pnorm(x<-1.5) # [1] 0.9331928 x # [1] 1.5 Ted.
I would regard modifying a variable within the parameters of a function call as pretty tasteless. What does: foo(x<-2,x) or foo(x,x<-3) do that couldn't be done clearer with two lines of code? Remember: 'eschew obfuscation'. Barry
Tasteless or not, the language allows it to be done; and therefore discussion of distinctions between ways of doing it is relevant to Erin's question! While I am at it, in addition to the above example, we can have x <- 1.234 sqrt(x=4) # [1] 2 x # [1] 1.234 compared with (as in the first example): x <- 1.234 sqrt(x<-4) # [1] 2 x # [1] 4 There is a passage in ?"<-" (which I don't completely understand) which is also relevant to Erin's query about '=' vs '<-': The operators '<-' and '=' assign into the environment in which they are evaluated. The operator '<-' can be used anywhere, whereas the operator '=' is only allowed at the top level (e.g., in the complete expression typed at the command prompt) or as one of the subexpressions in a braced list of expressions. (I'm not too clear about the scope of "one of the subexpressions in a braced list of expressions"). Ted. -------------------------------------------------------------------- E-Mail: (Ted Harding) <Ted.Harding at manchester.ac.uk> Fax-to-email: +44 (0)870 094 0861 Date: 15-Jan-10 Time: 09:08:41 ------------------------------ XFMail ------------------------------
On Fri, Jan 15, 2010 at 2:38 PM, Ted Harding
On 15-Jan-10 08:14:04, Barry Rowlingson wrote:
On Fri, Jan 15, 2010 at 6:57 AM, Ted Harding <Ted.Harding at manchester.ac.uk>wrote:
There is at least one context where the distinction must be preserved. Example: ?pnorm(1.5) ?# [1] 0.9331928 ?pnorm(x=1.5) ?# Error in pnorm(x = 1.5) : unused argument(s) (x = 1.5) ?pnorm(x<-1.5) ?# [1] 0.9331928 ?x ?# [1] 1.5 Ted.
I would regard modifying a variable within the parameters of a function call as pretty tasteless. What does: ?foo(x<-2,x) or ?foo(x,x<-3) do that couldn't be done clearer with two lines of code? ?Remember: 'eschew obfuscation'. Barry
Tasteless or not, the language allows it to be done; and therefore discussion of distinctions between ways of doing it is relevant to Erin's question! While I am at it, in addition to the above example, we can have ?x <- 1.234 ?sqrt(x=4) ?# [1] 2 ?x ?# [1] 1.234 compared with (as in the first example): ?x <- 1.234 ?sqrt(x<-4) ?# [1] 2 ?x ?# [1] 4 There is a passage in ?"<-" (which I don't completely understand) which is also relevant to Erin's query about '=' vs '<-': ?The operators '<-' and '=' assign into the environment in ?which they are evaluated. ?The operator '<-' can be used ?anywhere, whereas the operator '=' is only allowed at the ?top level (e.g., in the complete expression typed at the ?command prompt) or as one of the subexpressions in a braced ?list of expressions. (I'm not too clear about the scope of "one of the subexpressions in a braced list of expressions").
For example:
x = xyplot(1~1) system.time(x = xyplot(1~1))
Error in system.time(x = xyplot(1 ~ 1)) : unused argument(s) (x = xyplot(1 ~ 1))
system.time({ x = xyplot(1~1) })
user system elapsed 0.008 0.000 0.005 Of course, <- would not have had a problem. This is the most common problem I personally have had using = for assignment (better readability of <- is also a huge plus). -Deepayan
On Fri, Jan 15, 2010 at 2:38 PM, Ted Harding <Ted.Harding at manchester.ac.uk> wrote:
On 15-Jan-10 08:14:04, Barry Rowlingson wrote:
On Fri, Jan 15, 2010 at 6:57 AM, Ted Harding <Ted.Harding at manchester.ac.uk>wrote:
There is at least one context where the distinction must be preserved. Example: ?_pnorm(1.5) ?_# [1] 0.9331928 ?_pnorm(x=1.5) ?_# Error in pnorm(x = 1.5) : unused argument(s) (x = 1.5) ?_pnorm(x<-1.5) ?_# [1] 0.9331928 ?_x ?_# [1] 1.5 Ted.
I would regard modifying a variable within the parameters of a function call as pretty tasteless. What does: ?_foo(x<-2,x) or ?_foo(x,x<-3) do that couldn't be done clearer with two lines of code? ?_Remember: 'eschew obfuscation'. Barry
Tasteless or not, the language allows it to be done; and therefore discussion of distinctions between ways of doing it is relevant to Erin's question! While I am at it, in addition to the above example, we can have ?_x <- 1.234 ?_sqrt(x=4) ?_# [1] 2 ?_x ?_# [1] 1.234 compared with (as in the first example): ?_x <- 1.234 ?_sqrt(x<-4) ?_# [1] 2 ?_x ?_# [1] 4 There is a passage in ?"<-" (which I don't completely understand) which is also relevant to Erin's query about '=' vs '<-': ?_The operators '<-' and '=' assign into the environment in ?_which they are evaluated. ?_The operator '<-' can be used ?_anywhere, whereas the operator '=' is only allowed at the ?_top level (e.g., in the complete expression typed at the ?_command prompt) or as one of the subexpressions in a braced ?_list of expressions. (I'm not too clear about the scope of "one of the subexpressions in a braced list of expressions").
For example:
x = xyplot(1~1) system.time(x = xyplot(1~1))
Error in system.time(x = xyplot(1 ~ 1)) : unused argument(s) (x = xyplot(1 ~ 1))
system.time({ x = xyplot(1~1) })
user system elapsed 0.008 0.000 0.005 Of course, <- would not have had a problem. This is the most common problem I personally have had using = for assignment (better readability of <- is also a huge plus). -Deepayan
Thanks for spelling out "braced list", Deepayan. And I have to agree with you about the readability of "<-". Indeed, as summary advice to Erin, I wouild say use "<-" except in the specific context of assigning values to named parameters in functions, to named elements of lists, and the like (where use of "=" is mandatory anyway, or you won't get the desired effect -- see below). list(x<-1.234,y<-2.345) # [[1]] # [1] 1.234 # [[2]] # [1] 2.345 list(x=1.234,y=2.345) # $x # [1] 1.234 # $y # [1] 2.345 Ted. -------------------------------------------------------------------- E-Mail: (Ted Harding) <Ted.Harding at manchester.ac.uk> Fax-to-email: +44 (0)870 094 0861 Date: 15-Jan-10 Time: 10:01:55 ------------------------------ XFMail ------------------------------
G'day all, On Fri, 15 Jan 2010 09:08:44 -0000 (GMT)
On 15-Jan-10 08:14:04, Barry Rowlingson wrote:
[...]
I would regard modifying a variable within the parameters of a function call as pretty tasteless. What does: foo(x<-2,x) or foo(x,x<-3) do that couldn't be done clearer with two lines of code?
Depending on the definition of foo() more than two lines might be necessary to clarify what this code is actually supposed to do. :)
Remember: 'eschew obfuscation'.
Absolutely.
Tasteless or not, the language allows it to be done; and therefore discussion of distinctions between ways of doing it is relevant to Erin's question!
And a full discussion of these examples would include to warn about the
side effects of lazy evaluation:
R> rm(list=ls(all=TRUE))
R> foo <- function(x, y){
+ if(x > 0 ) x else x + y}
R> foo(2, x <- 10)
[1] 2
R> x
Error: object 'x' not found
R> foo(-2, x <- 10)
[1] 8
R> x
[1] 10
Having said this, I often use "plot(fm <- lm(y ~ . , somedata))".
And, yes, students have complained that the code I gave them was not
working, that they got a strange error message and they promised that
they had typed exactly what I had written on the lab sheet. On
checking, they had, of course, typed "plot(fm = lm(y ~ . , somedata))"
Cheers,
Berwin
On Fri, Jan 15, 2010 at 3:45 AM, Erin Hodgess <erinm.hodgess at gmail.com>wrote:
Hi R People: I'm teaching a statistical computing class using R starting next week (yay!) and I have an opinion type question, please. I'm old school and use "<-" in an assignment.
You call that 'old school'?? I still use ' x_1'! Of course ESS turns the underscore into '<-' magically. Perhaps these guys should redo their tests with slightly different syntax: http://www.cs.mdx.ac.uk/research/PhDArea/saeed/paper1.pdf Barry
I do like their experimental design:
"In a bizarre event which one of the authors insists was planned, and
the other maintains was a really stupid idea that just happened to work,
the test was ?rst administered to about 30 students on a...."
("Like" in the sense of being amused, of course)
-p
O__ ---- Peter Dalgaard ?ster Farimagsgade 5, Entr.B c/ /'_ --- Dept. of Biostatistics PO Box 2099, 1014 Cph. K (*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918 ~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk) FAX: (+45) 35327907
On Fri, Jan 15, 2010 at 6:57 AM, Ted Harding <Ted.Harding at manchester.ac.uk>wrote:
There is at least one context where the distinction must be
preserved. Example:
pnorm(1.5)
# [1] 0.9331928
pnorm(x=1.5)
# Error in pnorm(x = 1.5) : unused argument(s) (x = 1.5)
pnorm(x<-1.5)
# [1] 0.9331928
x
# [1] 1.5
Ted.
I would regard modifying a variable within the parameters of a function call as pretty tasteless. What does: foo(x<-2,x) or foo(x,x<-3) do that couldn't be done clearer with two lines of code?
The most common use I see that I like is within a conditional test like
if ( !is.null(x <- get("x", somehow)) && length(x) == 1) { dosomething }
The x variable is only used for the test, but since it is used twice
there, the assignment saves getting it twice. You could expand it to
two lines
x <- get("x", somehow)
if ( !is.null(x) && length(x) == 1) { dosomething }
but I find that a tiny bit harder to read.
On the other hand, I would never use the examples you gave, because I'd
have no idea what the value of x would be, since it depends on the order
of evaluation of the arguments. In R, I don't even know for sure if the
assignment would be evaluated at all, let alone before the x argument.
Duncan Murdoch
Remember: 'eschew obfuscation'. Barry
I've only been using R for about 2.5 years but and I'm not all that good but I vote for <- . I think the deciding factor is in RSiteSearch() and the various manuals. Almost everything I see uses <- . Why introduce = when it is not used normally? It will just confuse the students who are trying to use any of the documentation. Not to mention they might slammed for bad syntax on the R-help mailing list. :)
From: Erin Hodgess <erinm.hodgess at gmail.com> Subject: [R] advice/opinion on "<-" vs "=" in teaching R To: "R help" <r-help at stat.math.ethz.ch> Received: Thursday, January 14, 2010, 10:45 PM Hi R People: I'm teaching a statistical computing class using R starting next week (yay!) and I have an opinion type question, please. I'm old school and use "<-" in an assignment. However, I'm starting to see the "=" in the literature. Which should I use or does it matter, please? Thanks for your input! Sincerely, Erin -- Erin Hodgess Associate Professor Department of Computer and Mathematical Sciences University of Houston - Downtown mailto: erinm.hodgess at gmail.com
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
__________________________________________________________________ The new Internet Explorer? 8 - Faster, safer, easier. Optimized for Yahoo! Get it Now for Free! at http://downloads.yahoo.com/ca/internetexplorer/
John Kane <jrkrideau <at> yahoo.ca> writes:
I've only been using R for about 2.5 years but and I'm not all that good but
I vote for <- .
I think the deciding factor is in RSiteSearch() and the various manuals. Almost everything I see uses <- . Why introduce = when it is not used
normally? It will just confuse the
students who are trying to use any of the documentation. Not to mention they might slammed for bad syntax on the R-help mailing list. :)
Those are all good reasons. I have said something similar before (see <http://www.mail-archive.com/r-help at r-project.org/msg16904.html>), but I tend to use = because it seems to be more intuitive for students, despite being logically confused at a deeper level, and I want to spare them any additional cognitive load when they are first getting introduced to R. I'm not particularly convinced by the "<- is more general and there are some contexts where = doesn't work", because I'm not trying to be absolutely rigorous, nor teach all the possible ins and outs of R syntax. I would be very surprised if any of the examples given actually came up in the course of a first-semester statistics/modeling R course. I just want to do what works best for the students -- the problem is deciding on the balance between short term benefit (<- is one more odd thing to get used to) and long term benefit (they will see <- in other contexts, so they might as well get used to it eventually). Ben Bolker
On Fri, Jan 15, 2010 at 6:57 AM, Ted Harding <Ted.Harding at manchester.ac.uk>wrote:
There is at least one context where the distinction must be preserved. Example: pnorm(1.5) # [1] 0.9331928 pnorm(x=1.5) # Error in pnorm(x = 1.5) : unused argument(s) (x = 1.5) pnorm(x<-1.5) # [1] 0.9331928 x # [1] 1.5 Ted.
I would regard modifying a variable within the parameters of a function call as pretty tasteless. What does: foo(x<-2,x) or foo(x,x<-3) do that couldn't be done clearer with two lines of code?
It allows the lazy evaluation mechanism to determine whether the assignment happens. That would be very hard to do in two lines of code.
-thomas
Thomas Lumley Assoc. Professor, Biostatistics
tlumley at u.washington.edu University of Washington, Seattle
Hi R People: I'm teaching a statistical computing class using R starting next week (yay!) and I have an opinion type question, please. I'm old school and use "<-" in an assignment. However, I'm starting to see the "=" in the literature. Which should I use or does it matter, please?
I would say to use = if you are teaching people familiar with C or Java, and to use <- otherwise.
-thomas
Thomas Lumley Assoc. Professor, Biostatistics
tlumley at u.washington.edu University of Washington, Seattle
John Kane <jrkrideau <at> yahoo.ca> writes:
I've only been using R for about 2.5 years but and I'm not all that good ?but
I vote for <- .
I think the deciding factor is in ?RSiteSearch() and the various manuals. Almost everything I see uses <- . ?Why introduce = when it is not used
normally? ?It will just confuse the
students who are trying to use any of the documentation. Not to mention they might slammed for bad syntax on the R-help mailing list. ?:)
?Those are all good reasons. ?I have said something similar before (see <http://www.mail-archive.com/r-help at r-project.org/msg16904.html>), but I tend to use = because it seems to be more intuitive for students, despite being logically confused at a deeper level, and I want to spare them any additional cognitive load when they are first getting introduced to R. ? I'm not particularly convinced by the "<- is more general and there are some contexts where = doesn't work", because I'm not trying to be absolutely rigorous, nor teach all the possible ins and outs of R syntax. I would be very surprised if any of the examples given actually came up in the course of a first-semester statistics/modeling R course.
I teach the idiom summary(fm1 <- lm(y ~ x, mydata)) in my introductory courses.
I just want to do what works best for the students -- the problem is deciding on the balance between short term benefit (<- is one more odd thing to get used to) and long term benefit (they will see <- in other contexts, so they might as well get used to it eventually). ?Ben Bolker
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Barry Rowlingson wrote:
On Fri, Jan 15, 2010 at 3:45 AM, Erin Hodgess<erinm.hodgess at gmail.com>wrote:
Hi R People: I'm teaching a statistical computing class using R starting next week (yay!) and I have an opinion type question, please. I'm old school and use "<-" in an assignment.
You call that 'old school'?? I still use ' x_1'! Of course ESS turns the underscore into '<-' magically. Perhaps these guys should redo their tests with slightly different syntax: http://www.cs.mdx.ac.uk/research/PhDArea/saeed/paper1.pdf Barry
I do like their experimental design: "In a bizarre event which one of the authors insists was planned, and the other maintains was a really stupid idea that just happened to work, the test was ?rst administered to about 30 students on a...."
This is, of course, entirely consistent with the concept of outliers as potentially patentable discoveries that was discussed only a few days ago... Jim