| Summary: | 碩士 === 中華大學 === 營建管理學系碩士班 === 102 === As a fundamental industry, the construction industry can contribute to national economic development, and promote or affect development of other relevant industries. During early period of Japanese occupation, Taiwan developed transport construction to reduce transport costs and increase production capacity, which promoted Taiwan's economic growth. The economy was stimulated by increasing construction investment.
In recent years, the international energy crisis and rise of Southeast Asian countries have led to dramatic fluctuation of construction costs in Taiwan. As construction engineering requires large investment, long contract period and high risk, uncertain factors often occur, which may change construction cost price. If construction costs continuously rise, construction contractors may fail to survive, and cannot complete construction. Thus, it is urgent to construct an easy and accurate predictive model which can help enterprises make risk planning.
This study investigated construction cost indexes, bulk sale prices, customer price indexes, and export and import price indexes of 19 construction enterprises. The data were sourced from price indexes published by Directorate General of Budget, Accounting and Statistics, Executive Yuan, from January 2003 to July 2013. The period of each data was 127 months. ANOVA was used to analyze difference of correction indexes. In the principal component analysis, the construction price index was treated as the dependent variable, and variable attributes were integrated so as to reduce data. The scree diagram was plotted. The lowest threshold was an eigenvalue of greater than 1. A total of 2 principle components were extracted. According to the explanation, accumulated amount of variability of construction price index was 88.418%.
There were 19 variables. In order to solve multi-collinearity and establish an accurate prediction model, this study conduced stepwise multiple regression analysis. Forward method was used to select independent variable with the greatest amount of explanation variability in sequence, and backward method was used to eliminate selected variable with too small contribution. In stepwise multiple regression analysis, predictive power of the selected independent variables may reach significance level. The accuracy rate of the proposed regression equation reached 99% in prediction of indexes of next three months. The results can serve as reference for future investment and budgeting, and for construction companies in effective risk evaluation.
|
|---|
