Energy modeling of Hoeffding tree ensembles

Energy consumption reduction has been an increasing trend in machine learning over the past few years due to its socio-ecological importance. In new challenging areas such as edge computing, energy and predictive accuracy are key variables during algorithm design implementation. State-of-the-art ensemble stream mining algorithms able create highly accurate predictions at a substantial cost. This paper introduces nmin adaptation method ensembles of Hoeffding tree algorithms, further reduce their without sacrificing accuracy. We also present extensive theoretical models detailing patterns how affects consumption. have evaluated efficiency on five different trees under 11 publicly available datasets. The results show that we significantly, by 21% average, affecting less than one percent average.