A model for smoothing time-series data in construction

The understanding of the behaviour of time-series data has been a matter of concern to researchers and practitioners in a variety of fields ranging from social science and economics to engineering. Also, the behaviour of many phenomena within fields relating and peripheral to construction is described as a time series. Typically, the time-series analysis is carried out in order to forecast the future values of the series. These techniques, however, are also used to abstract the generalities within the series, hence facilitating the replication of the entire profile, reflecting only the main characteristics of the profile. There is a variety of techniques that can be applied to a set of time-related data. The choice of the technique is, therefore, dependent on the nature of the problem and the characteristics of the data. The diversity of available techniques is, on the one hand an advantage for all analysts. However, this diversity is also an indication that there is no universal technique that is applicable to a diversity of time-series data. This work fundamentally addresses the issue of smoothing and curve-fitting techniques rather than predicting and forecasting. A technique is offered which is tested against a set of criteria that are designed to focus on the accuracy of imitation and the practicality of operation: the ability to deal with a large number of time-series sets of data in a consistent, replicable and automated way. The viability of the technique is demonstrated by its application to expenditure profiles of a large number of construction projects. The size of the sample and the diversity in the profiles of the expenditure patterns provided an appropriate testing ground for the universality of the model. The results indicated that the model can effectively transform a jagged time series into a smooth pattern, while complying with a set of criteria many of which are common to several other research works relating to time-series data analysis. The proposed technique sequences a number of basic smoothing methods and the process involves the treatment and incorporation of the residual values.