PLOT Z(X,Y)=<function of X,Y> / TYPE=CONTOUR
(with DEVICE=PS or DEVICE=G or as GPLOT)
draws a contour plot of the given function of two variables as a
raster image.
The function is written according to the rules of curve plotting (see
CURVES?). X and Y may be replaced by any words. The scaling must be
indicated by the SCALE, XSCALE, YSCALE, X and Y specifications, for example,
as follows:
SCALE=0(0.1)1 X=0,1,0.01 Y=0,1,0.01
The last parameter (0.01) in X and Y specifications gives the steplength.
The ranges of XSCALE and X (and YSCALE and Y) must coincide.
The function is evaluated in the middle of each cell defined by the
X and Y specifications and these function values are mapped to various
colours or shades of gray.
In the color mapping the function values are assumed to be within the
interval (0,1). If they are not, the values are treated modulo 1.
This feature is very useful when clear contour curves should be obtained.
Before the color mapping, the function values f(x,y) can be linearly
tranformed to a*f(x,y)+b by giving a specification ZSCALING=a,b .
On the screen, various palettes may be used by the PALETTE specifi-
cation. It refers to a text file with extension .PAL.
For example, VGAGRAY.PAL gives 16 different shades of gray.
On PostScript devices, the true amount of gray shades greatly depends
on the resolution.
The SCREEN=NEG specification reverses the order of the palette colors
or gray shades. Default is SCREEN=POS.
On PostScript devices, the SCREEN specification has an extended form
SCREEN=<POS | NEG>,<raster_width_(samples/inch)>,<raster_angle> .
On the small laser printers (300 pixels/inch), the default setting is
SCREEN=POS,60,0 .
Example 1: Paraboloid
................................................................................
GPLOT Z(X,Y)=X*X+Y*Y / TYPE=CONTOUR
SCALE=-1(0.5)1 X=-1,1,0.05 Y=-1,1,0.05
ZSCALING=0.5,0
PALETTE=VGAGRAY
................................................................................
Example 2: Influence curves for correlation coefficient
CORR DECA,CUR+1 / VARS=Height,Weight
Means, std.devs and correlations of DECA N=48
Variable Mean Std.dev.
Height 186.9583 5.090493
Weight 85.56250 6.847600
Correlations:
Height Weight
Height 1.0000 0.8522
Weight 0.8522 1.0000
................................................................................
r=0.85 mx=186.96 my=85.56 sx=5.09 sy=6.85 n=48
HEADER=[MAGENTA],Influence_curves_for_correlation_coefficient
GPLOT z(x,y)=abs(r*(1-w)+u*v)/w
u=sqrt(n/(n*n-1))*(x-mx)/sx v=sqrt(n/(n*n-1))*(y-my)/sy w=sqrt((1+u*u)*(1+v*v))
x=150,220,0.25 y=40,130,0.25
TYPE=CONTOUR SCREEN=NEG 1/0.05=20
ZSCALING=20,0 (One band corresponds to 0.05 increase/decrease in corr.)
XSCALE=150(10)220 YSCALE=40(10)130 PALETTE=VGAGRAY
................................................................................
A general sucro for making these plots on the next page!
Sucro /INFL_R makes influence curves for the correlation coefficient
of two variables of any data set as contour plots.
The user may indicate a new point in the graph by the mouse and see
its effect to the correlation coefficient.
See
/INFL_R
Example:
/INFL_R DECA,Height,Weight,0.05
Reference: http://www.survo.fi/papers/corrcurves.pdf
P = More information on PLOT
Help System (web edition)