1987 Volume 22 Issue 6 Pages 416-424
A lot of measures which evaluate the performance of diffusion model have been suggested. It has not clalified what characteristics of the model these measures evaluate and the similarity among these measures.
It is necessary, however, to select the proper measures which suit the objectives of diffusion models and to judge the model in accordance with the quantitative standard.
In this paper, first the evaluation measures and correlation coefficients among them were derived based on the observed and the calculated concentrations using diffusion model which was applied to the Kashima areas. It was clear that the correlation coefficients among some measures were high value.
Next principal component analysis was introduced in order to make clear the characteristics of these evaluation measures toward the model. As a result, these measures were diveded into four groups, and then it was possible to evaluate easily the different ability of the diffusion model by means of the different four measures which were selected from each group.