Kitty Lee wrote: > Hi Roger, I forgot to mention I did try tripack and did go through your > previous post. > > But I'm doing a simulation and I have to repeat the process >1000 times. > > > I feel using the voronoi method to divide up the study area may not be the > most efficient way. I thought of dividing the area into grids(by generating > random x-coor and y-coor). But the tricky part is that the study region is > NOT rectangular. If I do 5*4=20 grids, since there're empty areas, I may have > fewer than 20 regions at the end..... > > > > I wonder how people divide their study area into a specific number of groups > when the study region is not rectangular? > > If you're discretizing the are with square grid cells, you may want to think about clustering methods on x and y coordinates, especially k-means where you control the number of groups. If you set the number of iterations to 0 and 1 and use random seeds (taken in the area), you may get random partitioning.
I still have trouble imagening why you would want random partitions, and what this random process should reflect. -- Edzer _______________________________________________ R-sig-Geo mailing list R-sig-Geo@stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/r-sig-geo
