A Method of Fuzzy-Theory-Based Computer-Aided Exploratory Data Analysis

When evaluating functional relationships from given data in an exploratory state of investigation, we only want to get an imagination of a possibly existing trend. Usually, methods of local or empirical regression are chosen to meet this purpose. In the paper we suggest an alternative using the possibilities of modern picture processing methods and computers. The data were transformed into a grey-tone picture representing a fuzzy observation set, the maximum trace of which yields the desired first impression of the trend and may serve as the starting point for further detailed evaluation of the functional relationship. The method is illustrated by a real-data numerical example.