Decision Trees and Rule Induction Applied to the Characterization of Oxidizing Activity of a Microbial Community in an Industrial Bioleaching Heap

In the context of a bacterial assisted leaching system, understanding the kinetics of microbial oxidation is essential to explore alternatives for process optimization. On a laboratory scale, there are empirical-analytic models that fit well to the dynamic of bacterial populations. However, at industrial level, the bioleaching process complexity together with the limited understanding of the relationship among biological, physicochemical and operational variables, have made their modeling elusive. In this paper we present the results of the characterization of the oxidizing activity of a microbial community in an industrial bioleaching operation. Based on an ad-hoc data mining process defined for this problem, decision trees are firstly constructed and then a set of rules are selected to characterize the oxidizing activity of a subset of strips in a bioleaching system. Following the same process of decision tree generation and then rule selection, a second subset of strips are characterized and results are compared between both subsets of bioleaching strips.