Gold mine dam levels and energy consumption classification using artificial intelligence methods

In this paper a comparison between two single classifier methods (support vector machine, artificial neural network) and two ensemble methods (bagging, and boosting) is applied to a real-world mining problem. The four methods are used to classify, thus monitoring underground dam levels and underground pumps energy consumption on a double-pump station deep gold in South Africa. In terms of misclassification error, the results show support vector machines (SVM) to be more efficient for classification of underground pumps energy consumption compared to artificial neural network (ANN), and surprisingly, to both bagging and boosting. However, in terms of other performance measures (i.e., mean absolute error, root mean square error, relative absolute error, and root relative squared error) artificial neural networks yield good results. In terms of underground dam level classification, SVM outperforms all the other methods with artificial neural networks (once again) having the best overall performance when other performance measures other than misclassification error are considered.