Skip to content

Matplot does not work with x being POSIXt class (PR#9412)

4 messages · Gregor Gorjanc, Peter Dalgaard, Don MacQueen

#
Hi,

Matplot works with x being Date class but not POSIXt. Here is the
example with R version 2.5.0 Under development (unstable) (2006-12-06
r40129)

Example:

x <- Sys.Date() - c(1:10)
y <- cbind(1:10, 10:1)
class(x)
## [1] "Date"
matplot(x, y)

x <- strptime(as.character(x), format="%Y-%m-%d")
## [1] "POSIXt"  "POSIXlt"
matplot(x, y)
Error in matplot(x, y) : 'x' and 'y' must have only 1 or the same number
of columns

Additionally, matplot with x being Date class does not use apropriate
annotation for x axis.

Thank you!

Gregor
#
gregor.gorjanc at bfro.uni-lj.si wrote:
Matplot plots matrices, it is not a bug that it doesn't work with other 
objects. Rather, it is  a coincidence that it works with dates when 
their internal representation is as numbers. 

It might be a useful enhancement, but things are not so easy: The root 
of the issue is that matrices of time/date objects don't behave nicely. 
Try, e.g.

structure(matrix(1:10, 2), class = "Date")
matrix(structure(1:10, class = "Date"), 2)

The former will actually work in the sense that it allows indexing, it 
just forgets about dimensions when printing. The latter throws away the 
class of x, and is used inside matplot() hence the annotation will not 
be that for a Date object.

Also, as.matrix.POSIXlt() does something completely different, so things 
get a bit tricky...
#
As Peter said, matplot plots matrices. The columns of a matrix are 
vectors of numbers. A POSIXlt object is not a vector of numbers, it 
is a list. So it shouldn't work. And should not be expected to work.

But with a POSIXct object it will work.

x <- strptime(as.character(x), format="%Y-%m-%d")
## [1] "POSIXt"  "POSIXlt"
matplot( as.POSIXct(x), y)
At 1:56 PM +0100 12/7/06, gregor.gorjanc at bfro.uni-lj.si wrote:
This is very easy to obtain:

# x is a Date object
matplot(x,y,xaxt='n')
axis.Date(1,x)

or

# x is a POSIXlt object
matplot( as.POSIXct(x), y,xaxt='n')
axis.POSIXct(1,x)

So easy, in fact, that I personally would not expect R core to spend 
time on it. One of the virtues of R is that the language is so rich 
that little tweaks like this are often very easy.
-Don
#
Hi,
Don MacQueen wrote:
...
Thank you Peter and Don for these insights. I see that changes would be
far from trivial/tricky and as Don pointed out simple tweaks by the user
can solve this.

Thank you and sorry for inconvenience!

Best wishes, Gregor