MCDA4ArcMap : An Open-Source Multi-Criteria Decision Analysis and Geovisualization Tool for ArcGIS 10

When faced with important decisions, we tend to base our decision-making on a rational framework, which often includes multiple decision criteria. Spatial decision problems have been characterized as a set of geographically defined decision alternatives (locations) with associated criterion values (e.g., Malczewski 1999). Within Geographic Information Systems (GIS), multi-criteria decision analyses (MCDA) tools have been used for decision support in environmental, transportation, and urban/regional planning, in waste management, as well as in hydrology, agriculture, and forestry, to name but a few areas of application (Malczewski 2006). Often, MCDA tools are only loosely coupled with GIS software (e.g., calculations completed in a spreadsheet) or take the form of custom implementations in a GIS scripting/programming environment. Few GIS vendors have integrated generic MCDA functionality in their products, with the notable exceptions of Idrisi’s Multi-Criteria Evaluation module and ArcGIS’ Overlay Toolset. Both of these operate on raster data layers using map algebra operations to combine cell values into an evaluation score for each candidate location (raster cell). In this technical note, we present “MCDA4ArcMap”, an open-source tool for MCDA and geovisualization of vector data in Arc-Map. The analytical functionality of the tool includes three MCDA methods: weighted linear combination (WLC), ordered weighted averaging (OWA), and a local variant of WLC (LWLC). WLC corre-sponds to the weighted overlay tool that readers may know from ArcGIS. As an extension of the criterion importance weighting in WLC, the OWA method allows the decision-maker to specify a de-gree of risk in their approach to decision-making. OWA has been implemented previously in Idrisi (Jiang & Eastman 2000). The recently proposed LWLC (Malczewski 2011) adjusts criterion impor-tance weights with regards to the local range of criterion values. Criterion weights are increased in a neighbourhood, if their local range is large relative to their global range in the study area, or decreased if the local range is relatively small. This approach ad-heres to the range-sensitivity principle that stipulates that criterion weights should depend on the ranges of criterion values occurring in a specific decision problem.