Decision Tree-Based Deterioration Model for Buried Wastewater Pipelines

Asset management provides a managerial decision-making framework for public agencies to monitor, evaluate, and make informed decisions about how to best maintain vital civil infrastructure assets. Among many steps required for implementing asset management, developing an accurate deterioration model is one of the key components because it helps infrastructure agencies predict remaining asset life. The accuracy of deterioration models highly depends on the quality of input data and the computational technique used in data analysis. Among many options of computational techniques, a decision tree offers the combination of visual representation and sound statistical background. The visual representation enables the decision maker to identify the relationship and interdependencies of each decision and formulate an appropriate prediction. This study developed a decision tree-based deterioration model for sewer pipes. The performance of the new model is then compared with conventional regression- and neural networks-based models that are also developed using the same data sets. The result shows that the decision tree outperformed other techniques in terms of accuracy (error rate). The paper also discusses different deterioration patterns of different categories of pipes.