Statistical Modeling in Absence of System Specific Data: Exploratory Empirical Analysis for Prediction of Water Main Breaks

Chen, Thomas Ying‐Jeh; Beekman, Jared; Guikema, Seth D.; Shashaani, Sara

doi:10.1061/(asce)is.1943-555x.0000482

Cited by 17 publications

(10 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The semi-parametric models are advantageous over parametric models since they are adaptable to the complexity of pipe failure data. Chen et al (2019) performed a comparison of models at two levels, the road level and the census level (2.3 km 2 grid areas), predicting the probability of failure using data recorded between 2010 and 2014. The study was one of the first to include a GAM with logistic regression function and found that it performed best compared to several other deterministic and machine learning models at the road level, being able to identifying areas of a high probability of failure more accurately.…”

Section: Logistic Regressionmentioning

confidence: 99%

“…The main disadvantage is the assumption that the variables are related to pipe failures in a linear nature, which fails to describe complex non-parametric relationships. Therefore, Chen et al (2019) expolred the logisitc approach using a GAM, which almost compared in performance to machine learning. However, other proabilistic models are more likely to outperform logisci regresion, such as Bayesian models.…”

Section: Commentsmentioning

confidence: 99%

See 1 more Smart Citation

An evolution of statistical pipe failure models for drinking water networks: a targeted review

et al. 2022

View full text Add to dashboard Cite

The use of statistical models to predict pipe failures has become an important tool for proactive management of drinking water networks. This targeted review provides an overview of the evolution of existing statistical models, grouped into three categories: deterministic, probabilistic and machine learning. The main advantage of deterministic models is simplicity and relative minimal data requirement. Deterministic models predicting failure rates for the network or large groups of pipes performs well and are useful for shorter prediction intervals that describe the influences of seasonality. Probabilistic models can accommodate randomness and are useful for predicting time to failure, interarrival times and the probability of failure. Probability models are useful for individual pipe models. Generally, machine learning describes large complex data more accurately and can improve predictions for individual pipe failure models yet are complex and require expert knowledge. Non-parametric models are better suited to the non-linear relationships between pipe failure variables. Census data and socio-economic data requires further research. The complexity of choosing the most appropriate statistical model requires careful consideration of the type of variables, prediction interval, spatial level, response type and level of inference is required.

show abstract

Section: Logistic Regressionmentioning

confidence: 99%

Section: Commentsmentioning

confidence: 99%

An evolution of statistical pipe failure models for drinking water networks: a targeted review

et al. 2022

View full text Add to dashboard Cite

show abstract

“…To further explore the second finding, we produce a rank-ordered capture plot of the nonclustered and each of the clustered scenarios in Figure 6. We refer the reader to works by Chen et al (2019) and for a full description of the methodology and a brief summary of the curve, and what it illustrates is as follows. When formulating a capital replacement plan, a utility often prioritizes based on risk.…”

Section: Resultsmentioning

confidence: 99%

“…It is beyond our focus to develop the most accurate prediction model, so we used insights derived from previous works regarding which model performs well in terms of risk prioritization. We implement a boosted trees classification model, since it was found to provide the most accurate prioritizations of high-risk assets based on previous studies (Chen et al, 2019;.…”

Section: Probability Of Failure Risk Analysismentioning

confidence: 99%

Optimizing cluster selections for the replacement planning of water distribution systems

2021

Self Cite

View full text Add to dashboard Cite

Risk assessment is an effective tool for revealing criticalities across an entire water system. However, standard outputs that report risk at the individual pipe segment level can have limited use for decision support since utilities typically plan for the replacement of aggregated areas to minimize disruption and cost. This research presents a process for aggregating individual pipes into high-risk clusters. Each cluster is a group of contiguous pipe which, when targeted, will reduce mobilization cost and community disruption. A graph search is first used to locate potential clusters, and then the optimal selection, which maximizes risk capture, is identified. An integer programming model is presented, and the process is implemented on a real system. Empirical trials suggest that the ability to prevent future breaks is not significantly reduced when prioritizing clusters rather than individual pipes. This shows that the proposed method can guide more cost-efficient planning for pipe replacements.

show abstract

“…Model scalability is perhaps the most complex challenge in pipe failure assessment, a theme widely discussed in the literature due to the gravitas of its inherent effect on predictive accuracy . As such, pre-processing steps are often required, including relocating of pipe failures, handling of missing and inaccurate data, and model scalability, employing pipe grouping methods at the network or census level (Chen et al 2019), k-means clustering (Kakoudakis et al 2018), or re-cutting pipes to improve model accuracy (Winkler et al 2018). Understanding these challenges is vital for both managers and professionals involved in developing pipe failure models, yet whilst literature has acknowledged some difficulties, to our knowledge, there are few studies explicitly discussing these challenges holistically.…”

mentioning

confidence: 99%

The challenges of predicting pipe failures in clean water networks: a view from current practice

2021

View full text Add to dashboard Cite

Pipe failure models can aid proactive management decisions and help target pipes in need of preventative repair or replacement. Yet, there are several uncertainties and challenges that arise when developing models, resulting in discord between failure predictions and those observed in the field. This paper aims to raise awareness of the main challenges, uncertainties, and potential advances discussed in key themes, supported by a series of semi-structured interviews undertaken with water professionals. The main discussion topics include data management, data limitations, pre-processing difficulties, model scalability and future opportunities and challenges. Improving data quality and quantity is key in improving pipe failure models. Technological advances in the collection of continuous real-time data from ubiquitous smart networks offer opportunities to improve data collection, whilst machine learning and data analytics methods offer a chance to improve future predictions. In some instances, technological approaches may provide better solutions to tackling short term proactive management. Yet, there remains an opportunity for pipe failure models to provide valuable insights for long-term rehabilitation and replacement planning.

show abstract

Statistical Modeling in Absence of System Specific Data: Exploratory Empirical Analysis for Prediction of Water Main Breaks

Cited by 17 publications

References 26 publications

An evolution of statistical pipe failure models for drinking water networks: a targeted review

An evolution of statistical pipe failure models for drinking water networks: a targeted review

Optimizing cluster selections for the replacement planning of water distribution systems

The challenges of predicting pipe failures in clean water networks: a view from current practice

Contact Info

Product

Resources

About