Hello. I have a multi-type point pattern, a ppp object in spatstat, with two types, cases and controls. About 2600 point altogether. Overall, about 2% are cases. All in a polygonal window. I'm interested in the spatially varying relative risk of being a case rather than a control. I'm aware of the relrisk() command in spatstat, and also the risk() command in the sparr package, both of which take a ppp object as an argument. I've done it with both. They yield essentially similar pixel images, except that the range of the relative risk, as shown in the colored legend, is very different in the two plots.
plot(relrisk(thin.ems.ppp, at = "pixels", weights = NULL, varcov = NULL, relative=TRUE, adjust=1, edge=TRUE, diggle=TRUE, se=FALSE, casecontrol=TRUE, case = "case")) yields relative risks ranging from 0 to about 0.16 In contrast, risk(thin.ems.ppp, log = FALSE, verbose = TRUE, doplot = TRUE) yields an image with relative risks ranging from 0 to about 40. I've read the documentation for the packages, but perhaps I am still misunderstanding what each package means by "relative risk." Can anyone comment? Thanks --Chris Ryan [[alternative HTML version deleted]] _______________________________________________ R-sig-Geo mailing list R-sig-Geo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-geo