Prediction of Pipe Failures in Wastewater Networks Using Random Forest Classification
Razieh Tavakoli, Ali Sharifara, Mohammad Najafi
Abstract
There are various methods to predict the wastewater pipe deterioration for the actual and future conditions of wastewater pipes. These methods can help decision-makers to plan for future inspection. Therefore, developing an accurate prediction of wastewater pipe deterioration is one of the key components as it helps infrastructure agencies to predict the remaining asset life. The main goal of this research is to develop a predictive model for wastewater pipes using random forest classification. The predictive models can efficiently predict wastewater status and can increase the certainty level of the current condition of wastewater pipes. The dataset of this study is obtained from the City of Los Angeles, which represents current wastewater information of the mainline sewer pipes. The developed random forest classification model has achieved a stratified test set false-negative rate, the false-positive rate, and an excellent area under the ROC curve of 0.81 in a case study application for the City of LA, California. An area under the ROC curve >0.80 indicates the developed model is an “excellent” choice for predicting the condition of pipes in a sewer network. The model developed is an essential decision support tool for wastewater utility managers in their prioritization of inspection, maintenance, and replacement.