DECISION SUPPORT ANALYSIS MODULES

: BAYES evaluates Bayes' Theorem. Multiple evidence maps (such as that produced by PCLASS ) are permitted as long as they are conditionally independent. Prior probabilities may be input in map form (and thus, like the evidence maps, may vary continuously over space). The user is also able to specify the confidence in the decision rule, I. e., the belief that the evidence supportive of the hypothesis is truly reflected in the evidence at hand -- p(e|e'). BAYES is an extension to what is sometimes called a Bayesian Weight of Evidence approach.

: Evaluates the fuzzy set membership values (possibilities) of data cells based on any of three membership functions: sigmoidal, j-shaped and linear. Monotonically increasing, monotonically decreasing, symmetric and asymmetric variants are supported. Other Fuzzy Set operations such as CON (concentration), DIL (dilation), AND and OR are covered by the standard modules TRANSFOR and OVERLAY .

: Produces an error matrix analysis of categorical map data compared to ground truth information. Tabulates error of omission and commission, marginal and total errors and selected confidence intervals. Per Category Kappa Index of Agreement figures are also provided. This module is a substatial revision and replacement for CONFUSE.

: Computes a best-fit set of weights by calculation of the principal eigenvector of a pairwise reciprocal comparison matrix in which each factor in a multi-criteria evaluation is compared to every other factor. Information on consensus and procedures for resolving lack of consensus are provided.

: Computes a Multi-Criteria Avaluation by means of a weighted linear combination of factors, subsequently masked by a set of constraints. Factor uncertainty information is propagated to the result to compute the uncertainty in the suitabilities derived.

: Rank orders the cells in a raster image. Ties may optionally be resolved by using the rank order of a second image. Both primary and secondary ranks may be in ascending or descending order. The procedure is used extensively in optimization problems such as with RECLASS for single objective decisions and MOLA for mulit-objective decisions.

: MDCHOICE is a multi-dimensional choice procedure that produces an output map indicating, for each cell, which of a series of input maps has the highest value. In cases where the input maps represent suitabilities for different objectives, the procedure provides one alternative to the multi-objective decision problem (although the MOLA procedure is preferable in most instances). In these cases, input maps should be standardized either with STANDARD or by means of the histogram equalization procedure in STRETCH .

: An iterative Multi-Objective Land Allocation routine. Input maps are ranked suitability maps such as would be produced by ranking (using RANK ) the output from a multi-criteria evaluation (using MCE ). The procedure uses a decision heuristic to resolve conflicts and is suitable for use with massive data sets.