Probabilistic Modeling in Design

Abstract

A general procedure is proposed for evolving the form of a density function that is consistent with the concept of subjective probability. The procedure directly applies new data information to the updating of the form of a density function without imposing on it any theoretical distribution that could restrict its shape, and permits the direct use of judgment arising from real world experience. It is based on the simple concept that sample size is a measure of confidence in the shape of a density function. Two possible algorithms are given, and the concept is extended for simple “true” or “false” events. The importance of probability in artificial intelligence is also dicusssed, and its essentially subjective nature is described. Procedures are briefly suggested.