You could create a graph using the following scripts which puts on a new graph the reference line and two ROC curve lines.
The values for the ROC curves are obtained by right clicking on the ROC curve graph you obtained in the analysis window and choosing "customize", then in the script window you get the script of the line. You could copy those and paste them in the script below replacing the lines that are currently their. repeat this for every ROC curv you want to add.
Hope this helps.
New Window( "ROC Curve",
Graph Box(
FrameSize( 600, 600 ),
X Scale( 0, 1 ),
Y Scale( 0, 1 ),
//add axis titles
xName( "1-Specificity" ),
yName( "Sensitivity" ),
//add reference line
Line Style( 0 );
Y Function( XTemp , XTemp );
//add first line
Line Style( 1 );
Text Size( 12 );
Text({0.45, 0.2}, "......... Group, BMI; AUC = 0.88");
pen size(2);
Line(
[0, 0, 0, 0, 0, 0, 0, 0.153846153846154, 0.230769230769231, 0.230769230769231,
0.307692307692308, 0.384615384615385, 0.384615384615385, 0.461538461538462,
0.461538461538462, 0.538461538461539, 0.615384615384615, 0.692307692307692,
0.769230769230769, 0.846153846153846, 0.923076923076923, 1, 1],
[0, 0.111111111111111, 0.222222222222222, 0.333333333333333, 0.444444444444444,
0.555555555555556, 0.666666666666667, 0.666666666666667, 0.666666666666667,
0.777777777777778, 0.777777777777778, 0.777777777777778, 0.888888888888889,
0.888888888888889, 1, 1, 1, 1, 1, 1, 1, 1, 1]
);
//add second line
Line Style( 4 );
Text Size( 12 );
Text( {0.45, 0.1}, "_.._.._.._ Group, BMI, Endo; AUC = 0.96" );
Line(
[0, 0, 0, 0, 0, 0, 0, 0.0833333333333333, 0.0833333333333333, 0.166666666666667,
0.25, 0.25, 0.333333333333333, 0.416666666666667, 0.5, 0.583333333333333,
0.666666666666667, 0.75, 0.833333333333333, 1, 1],
[0, 0.222222222222222, 0.333333333333333, 0.444444444444444, 0.555555555555556,
0.666666666666667, 0.777777777777778, 0.777777777777778, 0.888888888888889,
0.888888888888889, 0.888888888888889, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
);
);
);One way is to make use of the ROC Table which exists after you choose to make the ROC plot. Basically, for each test make the ROC Table into a data table, concatenate them using the option to create a new column identifying the source tables, then use Overlay Plot with the plots grouped by source table and turning on Overlay by Groups. You can do this interactively or with something like the following example which I pieced together by saving scripts after doing it interactively:
dt = open("$SAMPLE_DATA/Fitness.jmp");
fm = dt << Fit Model(
Y( :Sex ),
Effects( :Oxy ),
Personality( Nominal Logistic ),
Run Model(
Positive
Level( "M" ),
Likelihood Ratio Tests( 1 ),
Wald
Tests( 0 ),
ROC
Curve( 1 ),
Logistic
Plot( 1 ),
SendToReport(
Dispatch( {"Receiver Operating
Characteristic"}, "ROC Table", OutlineBox, Close( 0 ) )
)
)
);
fmReport = Report( fm );
fmReport["ROC Table"][Table Box( 1 )] << make into data
table;
data table(1) << set name("Oxy");
fm = dt << Fit Model(
Y( :Sex ),
Effects( :Runtime ),
Personality( Nominal Logistic ),
Run Model(
Positive
Level( "M" ),
Likelihood Ratio Tests( 1 ),
Wald
Tests( 0 ),
ROC
Curve( 1 ),
Logistic
Plot( 1 ),
SendToReport(
Dispatch( {"Receiver Operating
Characteristic"}, "ROC Table", OutlineBox, Close( 0 ) )
)
)
);
fmReport = Report( fm );
fmReport["ROC Table"][Table Box( 1 )] << make into data
table;
data table(1) << set name("Runtime");
Concatenate(
Data Table( "Runtime" ),
Data Table( "Oxy" ),
Output Table( "Both" ),
Create source column
);
Overlay Plot(
X( :Name( "1-Specificity" ) ),
Y( :Sensitivity ),
Grouping( :Source Table ),
Overlay Groups,
Step( 1 ),
Show Points( 0 ),
SendToReport(
Dispatch(
{},
"106",
ScaleBox( 2 ),
{Format( "Best", 10 ), Show Major Grid(
1 )}
),
Dispatch(
{}, "101", ScaleBox( 2 ), {Show Major Grid( 1 )} )
)
);
style="margin-bottom: 0.0001pt; line-height: normal;">
style="font-size: 10pt; font-family: "Courier New"; color: navy;">
style="font-size: 10pt; font-family: "Courier New"; color: black;">
