Perform an edge analysis on the STEPP GLM model estimate objects.
stepp.edge(est, criteria, j=2, boot=0, seed=17, showstatus=TRUE, debug=0)It returns the result of the identified edge with the STEPP subpopulations.
a STEPP estimate object.
criteria to be used to identify the cut point; abs or rel scale and by how much.
number of treatment, default to 2
perform a bootstrap analysis, default is none (0).
seed used for bootstrap, default is 17.
show the status of bootstrap, default is TRUE.
internal debug flag, default is 0
Wai-ki Yip
The criteria argument is a list with three elements:
trtid - the treatment id;
scale - "A" for absolute scale or "R" for relative scale;
threshold - amount in either absolute or relative scale that would consider a jump.
e.g. crit <- list(trtid=1, scale="A", threshold=-0.03)
Only support for STEPP GLM models for now. Bootstrap is computational time intensive but it will provide you with a quantification of the variability of the edge identified.
stwin