Time Series Analysis for Construction Productivity Experiments

Abstract

Time series analysis is a powerful statistical tool that can be of great value in evaluating the results of experiments to improve construction productivity. Time series analysis explicitly recognizes the importance of the order in which experimental data are observed, and the statistical dependence of observed data. This paper presents a brief overview of time series analysis and demonstrates its application using previously published data for a series of experiments involving crane lift cycle durations. In one experiment, it was shown that despite no apparent change in productivity, a new technology changed the nature of the work operation. In another experiment, a change in productivity was observed, but little change to the overall operation was seen. In a third experiment, a large change in the complexity of the operation can be seen as well as a clear learning effect. The underlying changes are due to differences in disturbances to the overall operation. These nonproductivity changes may be of great interest to the researcher, but they would not be identified using conventional productivity comparison techniques.