Abstract
In the management and the social system which contain human factors, there exist many cases that data including the human vagueness are treated. In this paper, we propose the new methods of estimation of discriminant functions in cases of one variable and two variables for the approximate maximum likelihood estimation by the fuzzy data obtained from two groups. In usual observations, it is desired that we get accurate data as much as possible, but in the case treating the data with human vagueness, it is more realistic that we consider the event which is observed by the interval data such as classified as fuzzy intervals. In this paper, therefore, we defined the fuzzy data using the concept of Zadeh's probability of fuzzy events and proposed new methods, employing this fuzzy data, to estimate the discriminant functions by usual statistical calculation. Furthermore, we investigated the availability of our method by computer simulation under realistic situations. Consequently, it became clear that our methods are useful.
