[prev in list] [next in list] [prev in thread] [next in thread] 

List:       r-sig-geo
Subject:    [R-sig-Geo] Determining changes in events counts along paths
From:       Aren Cambre <aren () arencambre ! com>
Date:       2012-11-29 2:15:29
Message-ID: CAA1mBrqvciDPhQ6j6i7eo6TYFmyf5Xr8em39AMX+3m3T48yX-g () mail ! gmail ! com
[Download RAW message or body]

I have a road network and a set of events that occurs on the road network.

I want to spot areas where there are significant changes in the amount of
these events over time.

I have proposed dividing the map into a grid and comparing event counts in
each square between successive time periods. For example, I could use a
grid with 1000 meter x 1000 meter squares and do a month-by-month
comparison.

This might be "good enough", but it still has two weaknesses:

   1. A significant change in events could be diluted if the "center of
   mass" of these events' locations or times lies near a grid boundary or near
   the time division.
   2. The grid size has to be a compromise between being large enough to
   catch clusters of events yet small enough to not include too many unrelated
   roadways.

I think a better approach would be one that traverses paths and is able to
organically spot these areas of changes over time. Another might be to
overlay an offset grid and offset time over the original data to overcome
the problem I identified in #1, although the end effect may not be much
different than smaller grid and time slices.

Any thoughts on this?

Aren

	[[alternative HTML version deleted]]

_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-geo
[prev in list] [next in list] [prev in thread] [next in thread]