A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Detecting Changes in Water Quality Data
Timely deployment of contaminant warning systems requires on‐line sensors and advancement of data analysis and decision support systems to accurately detect water quality changes. As a demonstration of event detection in water quality data, three water quality change‐detection algorithms were developed and used to detect changes in water quality observed at four locations within a distribution system. Each data set was “spiked” with simulated anomalous water quality values of 1 h duration and 10 levels of spike strength. The receiver operating characteristic (ROC) curve is proposed as an objective means of assessing and comparing water quality change‐detection algorithm results that is also able to provide the tradeoff between missed detections and false alarms on a single plot. The area under the ROC curve provides a measure of the sensitivity and specificity of each algorithm for each spike strength and location. Resulting areas under the ROC curve range from 0.46 for spike strengths of 1.0 (background) to 0.98 for strengths of 3.5 standard deviations from the mean, where an ROC curve area of 1.0 indicates perfect detection.
Detecting Changes in Water Quality Data
Timely deployment of contaminant warning systems requires on‐line sensors and advancement of data analysis and decision support systems to accurately detect water quality changes. As a demonstration of event detection in water quality data, three water quality change‐detection algorithms were developed and used to detect changes in water quality observed at four locations within a distribution system. Each data set was “spiked” with simulated anomalous water quality values of 1 h duration and 10 levels of spike strength. The receiver operating characteristic (ROC) curve is proposed as an objective means of assessing and comparing water quality change‐detection algorithm results that is also able to provide the tradeoff between missed detections and false alarms on a single plot. The area under the ROC curve provides a measure of the sensitivity and specificity of each algorithm for each spike strength and location. Resulting areas under the ROC curve range from 0.46 for spike strengths of 1.0 (background) to 0.98 for strengths of 3.5 standard deviations from the mean, where an ROC curve area of 1.0 indicates perfect detection.
Detecting Changes in Water Quality Data
McKenna, Sean A. (author) / Wilson, Mark (author) / Klise, Katherine A. (author)
Journal ‐ American Water Works Association ; 100 ; 74-85
2008-01-01
12 pages
Article (Journal)
Electronic Resource
English
Detecting changes in trends in water quality using modified Kendall's tau
| Online Contents | 2000
Online Contents | 2003
Detecting changes in mixed‐sampling rate data sequences
| Wiley | 2023