In probability and statistics we often have to estimate probabilities and parameters in probability distributions using a random sample. Instead of using a point estimate calculated from the data we propose using fuzzy numbers which are constructed from a set of confidence intervals. In probability calculations we apply constrained fuzzy arithmetic because probabilities must add to one. Fuzzy random variables have fuzzy distributions. A fuzzy normal random variable has the normal distribution with fuzzy number mean and variance. Applications are to queuing theory, Markov chains, inventory control, decision theory and reliability theory.
New method of dealing with imprecise probabilities, most of which not published before Includes supplementary material: sn.pub/extras
James J. Buckley
Extension control decision model decision problem decision theory fuzzy numbers fuzzy parameters fuzzy probability theory fuzzy sets probability theory uncertain probabilities