 |
 |
 |
|
 |
|
 |
|
|
 |
 |
 |
Decision heuristics
Let the available land of St. Gittal be 64 km2, represented in a raster model with 1 km2 resolution. The task shall now be to select 16 km2 of sheep pasture and 20 km2 of suitable wolf habitat. Since sheep and wolves share common needs they compete for at least some regions. Thus, it is very likely that the sets of raster cells, in which the sum of suitabilities is maximized for the single use, overlap considerably.
But how can the best of the many possible combinations of dividing up the conflicting cells be selected? There are computationally
expensive, mathematical means of comparing alternatives. Such choice functions often involve some form of optimization, i.e.
they require that each alternative be evaluated in turn. This is not practical for many realistic applications using large
raster data sets. However, this problem can be avoided using approximation procedures called choice heuristics instead of
choice functions. A 
heuristic is a procedure designed to solve a problem that ignores whether the solution is theoretically correct, but which in general
produces a good solution or solves a simpler problem that contains or intersects with the solution of the more complex problem.
Heuristics specify a procedure to be followed rather than a function to be evaluated. A heuristic is rather a "rule of thumb",
based on experience or experiments. The following animation shows such a heuristic procedure to allocate land in our use case.
 |
|
|
|
|
|
|
|
