Odp: FW: RES: bug in axis.Date? was (Re: newbie needs help plottingtimeseries)

Hi

Is it from chron package? As I do not have it available now I can 
not try it but the problem is not in plotting itself but in custom 
 x axis labeling when there are missing date on either end of  a series.

You have several options

Define fake time values for yet nonexisting data.
Do your labeling only for existing values.
Use xlim when calling your plot (which seems to me the most suitable)

compare

x<-1:10
y<-1:10
x[8:10]<-NA
y[8:10]<-NA
plot(x, y, axes=F)
box()
axis(1, at=1:10, labels=11:20)

with this

x<-seq(as.Date("2004-01-01"), as.Date("2008-06-01"),by="month")
x[50:54] <-NA
y <- seq(405,410,length=54)
y[50:54] <-NA
plot(x, y, axes=FALSE, ylim=c(405,410),col="red", pch=15,
type="o", ylab="metres above sea level", main="Main Plot Title")
axis(2, at=405:410, tck=1, col="gray60")
axis.Date(1, at=seq(as.Date("2004-01-01"), as.Date("2008-06-01"),
by="month"), labels=seq(as.Date("2004-01-01"), as.Date("2008-06-01"),
by="month"), tck=1, col="gray60")

which shall basically do the same. When you set explicit limits in 
plotting you can get rid of subsequent error in labeling axis.

plot(x, y, axes=FALSE, ylim=c(405,410),col="red", pch=15,
type="o", ylab="metres above sea level", main="Main Plot Title", 
xlim=c(as.Date("2004-01-01"), as.Date("2008-06-01")))

OK, It is not bug, but you has to go quite deep into axis.Date 
function to undestand what is going on and how to avoid such behaviour.

Regards
Petr

r-help-bounces at r-project.org napsal dne 02.07.2008 14:29:24:

Instead of 
x<-seq(as.Date("2004-01-01"), as.Date("2008-06-01"),by="month")

from=iso2chron("2004-01-01")
to= iso2chron("2008-06-01")

from <-chron2UTCsecs(from)
  to<-chron2UTCsecs(to) 

delta##how many secs i.e month time in secs

 xx <- seq(from=from, to=to, by=delta) 
    x <- UTCsecs2chron(xx)

I think it will works




-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On Behalf Of Leandro Marino
Sent: Wednesday, July 02, 2008 5:47 PM
To: Petr PIKAL; Kathi
Cc: r-help at stat.math.ethz.ch
Subject: [R] RES: bug in axis.Date? was (Re: newbie needs help
plottingtimeseries)

You can do this plot saying to R that your file is a time series with
the
ts() function.


serie=ts(x,....)
and then you use plot(serie)

-----Mensagem original-----
De: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]Em
nome de Petr PIKAL
Enviada em: quarta-feira, 2 de julho de 2008 05:04
Para: Kathi
Cc: r-help at stat.math.ethz.ch
Assunto: [R] bug in axis.Date? was (Re: newbie needs help plotting
timeseries)


Hi

"Kathi" <kathi at dropnet.ch> napsal dne 01.07.2008 18:48:47:

On Tue, 1 Jul 2008 14:33:12 +0200, Petr PIKAL wrote

"Kathi" <kathi at dropnet.ch> napsal dne 01.07.2008 13:39:18:

Thanks, Petr.

After fiddling around some more with my code I found out that this

seems

to

have something to do
with the number of records the table I'm reading (here: column

B112).

Somehow,

changing the
number of records for B112 changes the numbers in the error

message
-

though I

don't quite
understand how the number of records influences the axis intervals

or

their

labels, since I define
those explicitly...

Neither I. So I sent it to r-help maybe somebody else will know. If
you debug code of axis.Date you find that after some computing there
is this comparison where range is based on your input data (in your
case x), and z is from defined at sequence of dates.

z <- z[z >= range[1] & z <= range[2]]

and this comparison results in character(0), when plotting with this
artificial example.

plot(1:100, seq(405,410,length=100), axes=FALSE, ylim=c(405,410),
col="red", pch=15,
type="o", ylab="metres above sea level", main="Main Plot Title")
axis(2, at=405:410, tck=1, col="gray60")
axis.Date(1, at=seq(as.Date("2004-01-01"), as.Date("2008-06-01"),
by="month"), labels=seq(as.Date("2004-01-01"),

as.Date("2008-06-01"),

by="month"), tck=1, col="gray60")

I presume, that your x is from different interval of dates or it is
not a date.

My data comes from a groundwater monitoring program. Once a month the

water surface is

measured in some 170 stations, and I've been using R to visualize

these

measurements. In this
particular case, the water table ranges between approximately 406 and

408

metres above sea level,
hence the y axis interval between 405 and 410.
The values of my column x are dates of the format YYYY-MM-DD. If

something

were wrong there,
then I guess I would not be able to plot my data at all. But using the

limits

of 2004-01-01 and
2008-06-01, I can plot all the timelines for my 170 measuring stations

without

any problem. But
since it's now July, I wanted to shift the view by half a year. So I

used

"search and replace" in my code
to change to 2004-07-01 and 2008-12-01 (plus six months in both

cases).
That's

when I got the
error. I almost have a feeling that R doesn't like the "empty" part of

the

graph for July-December...  I
don't understand why this happens, but maybe it has something to do

with
the

comparison of ranges
you mentioned.

OK, It seems that you are right. Maybe you could add some expected x
date/time values so that x axis will have no NA values.

x<-seq(as.Date("2004-01-01"), as.Date("2008-06-01"),by="month")
y <- seq(405,410,length=54)
y[50:54] <-NA
plot(x, y, axes=FALSE, ylim=c(405,410),col="red", pch=15,
type="o", ylab="metres above sea level", main="Main Plot Title")
axis(2, at=405:410, tck=1, col="gray60")
axis.Date(1, at=seq(as.Date("2004-01-01"), as.Date("2008-06-01"),
by="month"), labels=seq(as.Date("2004-01-01"), as.Date("2008-06-01"),
by="month"), tck=1, col="gray60")

as you see, then everything works. However I think that it is pretty
close
to bug in axis.Date, maybe somebody can enlighten us both :-)

Regards
Petr

Thanks for all your help!

Kathi

Regards
Petr

Kathi


On Mon, 30 Jun 2008 11:48:43 +0200, Petr PIKAL wrote

Hi

r-help-bounces at r-project.org napsal dne 30.06.2008 11:14:48:

Dear R help list

I have recently started using R to visualize time series of

water

level

measurements. The code I wrote
does exactly what I want:

plot(x, a$B112, axes=FALSE, ylim=c(405,410), col="red",

pch=15,

type="o", ylab="metres above sea level", main="Main Plot

Title")

axis(2, at=405:410, tck=1, col="gray60")
axis.Date(1, at=seq(as.Date("2004-01-01"),

as.Date("2008-06-01"),

by="month"), labels=seq(as.Date("2004-01-01"),

as.Date("2008-06-01"),

by="month"), tck=1, col="gray60")

Now I wanted to shift the data visualized by 6 months, so I

replaced

"2004-01-01" by "2004-07-01"
and "2008-06-01" by "2008-12-01". Instead of the desired plot,

I
get

an

error

message stating
(forgive the perhaps faulty translation into English, I'm

working on

a

German

version of R):

mistake in axis(side, at = z, labels = labels, ...) :
  length of 'at' and 'labels' different, 50 != 54

Those two calls have different length

at=seq(as.Date("2004-01-01"), as.Date("2008-06-01"), by="month")
labels=seq(as.Date("2004-01-01"), as.Date("2008-06-01"),

by="month")

you can try it by

length(seq(...))

could not there be some typo in your call?

Regards

Petr

What have I done wrong? Thank you very much for your help!

Katharina Dubach

code.

--
DropNet AG - Das Unternehmen fuer Ihren Internet-Auftritt!

--
DropNet AG - Das Unternehmen fuer Ihren Internet-Auftritt!