[R] Problem adding lines to a plot, when y is defined

[David Winsemius]

On Sep 9, 2013, at 12:27 PM, Cech, Christian wrote:

>>> On Sep 9, 2013, at 9:00 AM, Cech, Christian wrote:
>>> 
>>>> Dear all,
>>>> 
>>>> I want to create a line-plot with two lines and some additional scatter-plots.
>>>> However, adding a line to the plot does not work when I specify the y-argument in the plot command, while it does work when y is not specified.
>>>> 
>>>> I first send you an example that works:
>>>> plot(var[, 3],
>>>>   type="l",
>>>>   ylim=c(-0.04, 0),
>>>>   ylab   = 'portfolio returns',
>>>>   xlab   = 'time')
>>>> lines(var[, 5], type="l", lty=3)
>>>> for (i in 1:nrow(var)) {
>>>> if(var[i, 4] | var[i,6])
>>>>  points(i, var[i, 2], pch=4)
>>>> }
>>>> 
>>>> --> the result is displayed in attachment "Plot1.pdf"
>>>> 
>>>> What does not work is the following code, where as the first argument of plot (ie the y-argument) is defined:
>>>> plot(as.Date(var[, 1], origin="1899-12-30"),
>>>>   var[, 3],
>>>>   type="l",
>>>>   ylim=c(-0.04, 0),
>>>>   ylab='portfolio returns',
>>>>   xlab='time')
>>>> lines(var[, 5], type="l", lty=3)
>>>> for (i in 1:nrow(var)) {
>>>> if(var[i, 4] | var[i,6])
>>>>  points(i, var[i, 2], pch=4)
>>>> }
>>>> 
>>>> --> the result is displayed in attachment "Plot2.pdf"
>>>> 
>>>> The same problem appears if instead of
>>>> 
>>>> as.Date(var[, 1], origin="1899-12-30")
>>>> 
>>>> I use
>>>> 
>>>> var[, 1]
>>>> 
>>>> to define the y-argument.
>>> 
>>> This puzzles me. At the moment yu are using postional matching and it is var[,3] that would be matched to the y argument.
>>> 
>>> You need to post output of str(var) so that we can see the class of var[,1].
>>> 
>>> 
> 
>> Dear David,
>> 
>> thank you for the quick reply! Unfortunately I confused you because I mixed up y-value and x-value. What I actually wanted to say is that the plot does not work when x is specified (2nd example) while it does work if x is not specified (1st example). Sorry for this!
>> y = var[, 3]
>> x = var[, 1]
>> 
>> When I type str(var) i get the following output:
>> num [1:4747, 1:6] 33450 33451 33452 33455 33456 ...
>> 
>> Is that the information you needed?
> 
> Maybe. 'var' is  a numeric matrix which may not be what you think it was if it started out life as having dates. (On this list it is requested that you post in context which means bottom posting in many cases. So I moved your reply.)
> 
> When you apply the as.Date function with an origin, you will get an different number as the internal representation of the Date value than what you put in:
> 
> as.numeric(as.Date(c(33450, 33451, 33452, 33455, 33456), origin="1899-01-01"))
> [1] 7518 7519 7520 7523 7524
> I suspect that you have a misregistration of the x values. Since it is a matrix we might have better view of it with:
> 
> dput(head(var))
> 
> I also supect that you will end up using the `axis` function to do your labeling.
> 
> --
> 
> David Winsemius
> Alameda, CA, USA
> 
>> dput(head(var))
> structure(c(33450, 33451, 33452, 33455, 33456, 33457, -0.00299369609998333,
> -0.000882805884818571, -0.000570821462061429, 0.00735101809395667,
> -0.000574642535598095, -0.000464474459045714, -0.010510137214196,
> -0.0105173750100507, -0.0104987866171353, -0.0104808796029616,
> -0.0104561013745608, -0.0104562836793838, 0, 0, 0, 0, 0, 0, -0.0115901618181903,
> -0.0115901618181903, -0.0115901618181903, -0.0115901618181903,
> -0.0115901618181903, -0.0115901618181903, 0, 0, 0, 0, 0, 0), .Dim = c(6L,
> 6L), .Dimnames = list(NULL, c("col1", "col2", "col3", "col4",
> "col5", "col6")))
> 
> 

You need to get the x-axis values registered corectly for `lines` and `points` calls:

plot(as.Date(var[, 1], origin="1899-12-30"),
  var[, 3],
  type="l",
  ylim=c(-0.04, 0),
  ylab='portfolio returns',
  xlab='time')
dts <- as.numeric(as.Date(var[, 1], origin="1899-12-30"));lines(dts, var[, 5], type="l", lty=3)
for (i in 1:nrow(var)) {
if(var[i, 4] | var[i,6])
 points(dts[i], var[i, 2], pch=4)

-- 

David Winsemius
Alameda, CA, USA