STATISTICAL ANALYSIS OF TEXAS STATE HIGHWAYS' COST INDICES

摘要

Estimating the cost of a highway project is of paramount importance for contractors, managers, and clients. However, in a larger scale, departments of transportation need estimates of total costs of their highway projects for which they have developed cost indices. The objective of this study is applying normal probability plot, analysis of variance, hypothesis testing, regression analysis, Monte Carlo simulation, and confidence interval to analyze the indices published by the Texas Department of Transportation. The results indicate that one-month moving average unit bid prices, as well as the category indices, follow a normal distribution. Furthermore, at the significance level of 5%, surface course and pavement type, as well as the year, affect the unit bid price, while month does not. Additionally, regarding effects of the studied factors on the unit bid price, an interaction exists between all combinations of the month, year, and surface course or pavement type except surface course or pavement type and month. In addition, the type of work-item and year affect the work-item category indices, while month does not. Moreover, a verified linear regression was fitted to the work-item category indices. Finally, Monte Carlo simulation which was applied to determine 95% confidence interval for the mean of earthwork category index was found to produce an accurate enough confidence interval.