Support tools to predict the critical structural condition of uninspected sewer pipes in Bogota DC

Andres Eduardo Torres Abello, Nathalie Hernández, Nicolas Caradot, Hauke Sonnenberg, Pascale Rouault

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The objective of this work is to explore five methods based on learning machine and statistical approaches (logistic regression –LR–, random forest –RF–, multinomial logistic regression –Multi_LR, lineal discriminant analysis –LDA– and support vector machine –SVM–) considering: (i) only the age as independent input variable and (ii) other sewer characteristics together with age as input variables in the models in order to establish which one is the most adapted as sewer asset management support tool for CCTV dataset from Bogota. From ROC space and performance curves techniques, it is possible to establish that LR and RF methods are suitable to predict the structural condition of the uninspected sewer pipes and to identify pipes in critical conditions, considering more than one variable as the independent input variables in the models.
Original languageEnglish
Title of host publicationThe leading edge sustainable asset management of water and wastewater infrastructure conference
StatePublished - 2017

Cite this