| Summary: | 碩士 === 國立高雄第一科技大學 === 營建工程所 === 98 === Predicting Project Duration from Project Cost
-Using Second Freeway Construction Projects as Sample
Abstract
Being capable of rapid and accurate prediction of project duration has always been the common goal of owners and contractors. For a project to achieve success, planning and design in the early phases have a greater importance than tendering, construction, and maintenance in the later phases. In financial planning for a public construction project, it is necessary to allocate the budget for each year, and hence it is necessary to estimate reasonable project duration as the basis for distributing the funds for each year. Subsequently, during tendering, because the time available for bid preparation is usually insufficient for detailed time estimation, contractors need to have the ability to check quickly the correctness of project duration stipulated by the owner, in order to avoid the difficulty in project execution later on or the loss caused by the penalty for passing the contract deadline.
In light of the above, this study aimed to develop a model for predicting the duration of a project. First, 110 cases of completed projects for construction of the Second Freeway were collected, and the actual cost and the actual time of each case were adjusted for obtaining reasonable quantities for use as sample data. Next, the regression method and the back-propagation neural network were used separately to develop models for predicting project duration from project cost, while their errors in prediction were tested. The results show that the regression model built using the time and cost logarithms as proposed by Browmilow and the non-linear model built using the back propagation neural network both attained a mean absolute percentage error of around 25% in testing of duration prediction. Provided that further improvements are made, these two models would have the potential for a wider use in predicting reasonable duration for similar projects in Taiwan.
|
|---|
