DECISION-TREE CONSTRUCTION AND ANALYSIS

Decision-tree construction and analysis indicate the quantitative occu rrence of sulfate-reducing bacteria in a groundwater basis. The decisi on-tree analysis allowed determination of the most important measured variables (discriminators) associated with the occurrence of sulfate-r educing bacteria. Those variables may be used to improve reliability a nd validity of the tree. Statistical validation of endpoint data (leav es) is described, and decision trees are presented that increase the u nderstanding of individual wells and hypothesize causes for sulfate-re ducing bacteria.