1985 年 1985 巻 353 号 p. 67-74
As the computer has been developed, increased in size and become generally available, so Quantification Methods have been applied to various fields, e. g. individual behavior modeling and urban environment evaluation. But there are still many problems and some of their applications are unreliable for the practical purpose.
The aim of this study is to investigate the reliability of Quantification Method Type II with a special attention to its applicability for Infrastructure Planning. Through a Monte Carlo Simulation experiment, we have made it clear that the reliability is largely determined by three major factors, i. e. the number of samples, correlation ratio and the number of categories. Among them, the number of samples is the most important factor. When the number of samples is insufficient, the results will not be in agreement with the true values.