Predicting sewer structural condition using hybrid machine learning algorithms

Predicting the structural condition of sewer pipes plays a vital role in predictive maintenance and renewal plans many water utilities. This study explores simultaneous utilization physical environmental features prediction. Three (3) hybrid machine learning models which are combination Bagging (BG), Dagging (DG), Rotation Forest (RotF) ensembles with J48 Decision Tree (J48DT) based classifier were used to predict pipe conditions Ålesund city, Norway. The classification performance was evaluated using area under receiver operating characteristic (AUC-ROC) precision-recall (AUC-PRC) curves. RotF-J48DT model had highest (AUC-ROC = 0.857, AUC-PRC 0.918) values, followed by BG-J48DT, base J48DT. should be considered when predicting area.