Oystein wrote
I have stratified my data in four groups and are making Roc-curves for each group in SPSS. Is it somehow possible to get these four curves in one figure?
I have done something similar (i.e., ROC curves for different tests in the same plot) by using the following approach:
OMS
/SELECT TABLES
/IF COMMANDS = ['ROC Curve']
SUBTYPES = ['Coordinates of the Curve']
/DESTINATION FORMAT = SAV NUMBERED = group viewer = yes
OUTFILE = { file name for coordinates of the curve }.
OMS
/SELECT TABLES
/IF COMMANDS = ['ROC Curve']
SUBTYPES = ['Area Under the Curve']
/DESTINATION FORMAT = SAV NUMBERED = group viewer = yes
OUTFILE = { file name for AUC table } .
* Now run ROC for each group separately,
* e.g., via a series of filters, or SPLIT FILE by GROUP.
OMSEND.
* ---------------------------------------------- .
* Open the file of ROC data and generate the plot .
* ---------------------------------------------- .
get file = { file name for coordinates of the curve } .
var lab group "Group" .
val lab group
1 'Group 1 label'
2 'Group 2 label'
.
freq group.
formats @1Specificity Sensitivity (f8.2).
* Plot coordinates in a scatter-plot .
GRAPH
/SCATTERPLOT(BIVAR)=@1Specificity WITH Sensitivity BY model
/MISSING=LISTWISE
/TITLE= 'ROC Curve'.
Then I used the chart editor to add interpolation lines and do some other formatting. There is probably a way to add the interpolation lines via the GGRAPH version of a scatter-plot, but I didn't take the time to figure it out.
The reason for the other OMS command above is that I also wanted a table that summarized the AUC values for all of the curves in the plot.
HTH.
--
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/"When all else fails, RTFM."
PLEASE NOTE THE FOLLOWING:
1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above.
2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (
https://listserv.uga.edu/).