| Summary: | 碩士 === 正修科技大學 === 營建工程研究所 === 98 === In construction industry, a scheduled construction project is often affected by the failure to accurately grasp the trend of price fluctuation at the stage of material procurement and contract out; interface problem is a good example in this regard. Based on this observation, the present study aims to establish a technical indicator forecast model for Construction Cost Index in an attempt to assist the industry in appropriately deciding the time point of construction contract out and effectively control the cost and starting the interface coordination mechanism ahead of time.
In order to establish an effective technical index, data relating to economic index are collected from January, 1991 to December, 2009 in this study. After analysis, five major indices, including Taiwan Stock Exchange Capitalization Weighted Stock Index, Leading Indicator, Taiwan Stock Index Futures, Iron and Steel Wholesale Price Index and International Crude Oil Price are selected. According to the characteristics of these five indices, this study attempts to establish a technical indicator forecast model with more accurate prediction capability.
To effectively verify the model’s forecast capability, this study conducted a forecast to the samples of Construction Cost Index collected from July, 2008 to December, 2009. The result shows that this model’s forecast capability is relatively weak when goods price is affected by short-term fluctuation (for instance, at the last phase of financial meltdown). Therefore, the first half of forecast duration shows lesser stability in which the deviation value of forecast ranges from 0.1% to 5.22% and the mean deviation equals 3.43%; by contract, since goods price at the second half of forecast duration is interfered by lesser non-economic factors, the deviation value of forecast ranges from 0.19% to 2.26% and the mean deviation is 0.86%. The result of verification suggests that this model can effectively forecasts the trend of goods price fluctuation for four months later.
|
|---|
