Level: Intermediate
TCM Section(s)
6.4. Forensic Performance Assessment
9.2. Progress and Performance Measurement
Venue: 2024 AACE International Conference & Expo
Abstract: The measured mile method of estimating productivity loss uses actual project data to compare periods of expected productivity to periods of impacted productivity. The difference between expected and impacted productivity is an evaluation of the productivity loss that, when properly tied to an analysis of cause-and-effect, becomes the basis for determining the quantitative impact of those cause(s) and effect(s) on productivity.
A measured mile analysis requires identification of a reliable baseline to provide insight into the contractor’s unimpacted performance as it relates to productivity (hours/unit of measure). When evaluating actual productivity, the period of impacted performance must be identified and isolated to the periods of work impacted by the cause(s) and effect(s) being considered. While this approach is simple in theory, one of the practical challenges encountered is that the available productivity data, for any number of reasons, may include data that is not representative of the actual effort involved or the actual physical work accomplished within a given period. In other words, the information relied upon can be characterized as “noisy” data. Basing an analysis on noisy data can result in inaccurate and unreliable evaluations of productivity loss.
This paper highlights three practical approaches to dealing with noisy productivity data. The first approach provides a method for identifying and discounting ramp-up and ramp-down periods of actual productivity as well as offering recommendations for assessing the robustness of the data sets used in the analysis to assure representative data sets have been selected. The second approach employs rolling averages to analyze achieved productivity rather than the “all-inclusive” averages typically employed. The third approach presents the advantages of using statistical models to identify and treat anomalies and outliers in productivity data. These practical approaches can enhance the accuracy of analyzing productivity impacts using a measured mile.
TCM Section(s)
6.4. Forensic Performance Assessment
9.2. Progress and Performance Measurement
Venue: 2024 AACE International Conference & Expo
Abstract: The measured mile method of estimating productivity loss uses actual project data to compare periods of expected productivity to periods of impacted productivity. The difference between expected and impacted productivity is an evaluation of the productivity loss that, when properly tied to an analysis of cause-and-effect, becomes the basis for determining the quantitative impact of those cause(s) and effect(s) on productivity.
A measured mile analysis requires identification of a reliable baseline to provide insight into the contractor’s unimpacted performance as it relates to productivity (hours/unit of measure). When evaluating actual productivity, the period of impacted performance must be identified and isolated to the periods of work impacted by the cause(s) and effect(s) being considered. While this approach is simple in theory, one of the practical challenges encountered is that the available productivity data, for any number of reasons, may include data that is not representative of the actual effort involved or the actual physical work accomplished within a given period. In other words, the information relied upon can be characterized as “noisy” data. Basing an analysis on noisy data can result in inaccurate and unreliable evaluations of productivity loss.
This paper highlights three practical approaches to dealing with noisy productivity data. The first approach provides a method for identifying and discounting ramp-up and ramp-down periods of actual productivity as well as offering recommendations for assessing the robustness of the data sets used in the analysis to assure representative data sets have been selected. The second approach employs rolling averages to analyze achieved productivity rather than the “all-inclusive” averages typically employed. The third approach presents the advantages of using statistical models to identify and treat anomalies and outliers in productivity data. These practical approaches can enhance the accuracy of analyzing productivity impacts using a measured mile.