The objective of this study is to analyze and propose machine learning models to predict bursting and leaks in water connections since a significant portion of water leakage occurs in those connections. The following methods were used to predict and classify water connections vulnerability to ruptures and leaks: Logistic Regression, Knears Neighbors, Support Vector Machine, Decision Tree Classifier, Naive Bayes, Random Forest, AdaBoost, Gradient Boosting, and XgBoost. Data used in this study were collected from water supply systems in Brasília, capital of Brazil. Altogether, the water distribution system investigated is composed of 543,168 water connections as well as of 5.3 thousand kilometers of pipes of different materials such as high-density polyethylene, cast iron, and polyvinyl chloride. The infrastructure has been implemented since the 60's. The historical database on ruptures and leaks in water connections covers the period between January 2015 and April 2020. Data regarding possible factors was used such as material, diameter, network implantation year, water connection installation year, type of road traffic over the networks and water connections, and operational pressures of the water distribution network. Built models could assist the distribution network management process by identifying areas that can be assigned to leaks search campaigns or infrastructure rehabilitation. The intervening factor were analyzed to comprehend the influence of each component in deterioration process on the case study. In addition, performance obtained by different machine learning methods were compared.

Access the Conference Proceedings Paper on ASCE Library

The objective of this study is to analyze and propose machine learning models to predict bursting and leaks in water connections since a significant portion of water leakage occurs in those connections. The following methods were used to predict and classify water connections vulnerability to ruptures and leaks: Logistic Regression, Knears Neighbors, Support Vector Machine, Decision Tree Classifier, Naive Bayes, Random Forest, AdaBoost, Gradient Boosting, and XgBoost. Data used in this study were collected from water supply systems in Brasília, capital of Brazil. Altogether, the water distribution system investigated is composed of 543,168 water connections as well as of 5.3 thousand kilometers of pipes of different materials such as high-density polyethylene, cast iron, and polyvinyl chloride. The infrastructure has been implemented since the 60's. The historical database on ruptures and leaks in water connections covers the period between January 2015 and April 2020. Data regarding possible factors was used such as material, diameter, network implantation year, water connection installation year, type of road traffic over the networks and water connections, and operational pressures of the water distribution network. Built models could assist the distribution network management process by identifying areas that can be assigned to leaks search campaigns or infrastructure rehabilitation. The intervening factor were analyzed to comprehend the influence of each component in deterioration process on the case study. In addition, performance obtained by different machine learning methods were compared.

**[Access the Conference Proceedings Paper on ASCE Library](https://doi.org/10.1061/9780784483466.093)**

Water Connection Bursting and Leaks Prediction Using Machine Learning

Created in 1999, the Environmental & Water Resources Institute (EWRI) is the recognized leader within ASCE for the integration of technical expertise and public policy in the planning, design, construction, and operation of environmentally sound and sustainable infrastructure impacting air, land and water resources.

[Paper Proceedings](https://doi.org/10.1061/9780784483466)

Conference DOI: https://doi.org/10.26226/m.6329c39b68fb0e00131f77d2