Knowledge discovery by means of inductive methods in wastewater treatment plant data

Artificial intelligence techniques, including machine learning methods, and statistical techniques have shown promising results as decision support to ols, because of their capabilities of knowledge discovery, heuristic reason ing and working with uncertain and qualitative information. Wastewater trea tment plants are complex environmental processes that are difficult to mana ge and control. This paper discusses the qualitative and quantitative perfo rmance of several machine learning and statistical methods to discover know ledge patterns in data. The methods are tested and compared on a wastewater treatment data set. The methods used are: induction of decision trees, two different techniques of rule induction and two memory-based learning metho ds: instance-based learning and case-based learning. All the knowledge patt erns discovered by the different methods are compared in terms of predictiv e accuracy, the number of attributes and examples used, and the meaningful- ness to domain experts.