[R] Interesting Behavior in plot()

[Duncan Murdoch]

On Mon, 29 Mar 2004 14:06:36 -0800 (PST), Clint Bowman
<clint at ecy.wa.gov> wrote :

>I have a 2 by 226200 table, conveniently read in by read.table(), which
>exhibits some strange behavior when plotted by plot(V1,V2).  The general
>pattern for the range of windspeeds, [0<V1<50] is as expected -- the wind
>gust falls in the interval [V1<V2<65] except for certain values of V2.  
>For V2 == c(15,26,37,48,59), the V2 values are positioned at one-tenth of 
>the V1 (i.e., as if I had issued the command plot(0.1*V1,V2) for just 
>those values of V2 -- see attached plot, test.png).  The only other values 
>of V2 in the range, 4 and 70, also seem to be affected but aren't well 
>enough represented to show up clearly on the plot.
>
>However, if I plot(ws$V1[ws$V2==48],ws$V2[ws$V2==48]), I see the second 
>attachment, test2.png, which confirms that the wind speed (V1) really 
>should be positioned where one would expect.  I haven't seen any messages 
>covering this behaviour and so am looking for an explanation.

I can't quite see how it would generate the symptoms you saw, but a
common source of oddities is that data is read as factors rather than
as numeric values.  You can see the difference using the str()
function, e.g.

> str(x)
 num [1:10] -0.4897  0.6804  0.6979 -0.1203 -0.0428 ...
> str(as.factor(x))
 Factor w/ 10 levels "-1.65972758..",..: 4 7 8 5 6 9 2 10 3 1

I'd check your V1 and V2.

Duncan Murdoch