Fuzzy Weighted Average: Alternative approach

We present an alternative definition of the fuzzy weighted average, in which Zadeh's extension principle is applied to the definition of the non-fuzzy weighted average where weights are required to be normalised. It is argued that the alternative approach should be preferred above the traditional approach. An algorithm for the computation of the fuzzy weighted average for the alternative approach is given for the case where attributes and weights are fuzzy triangular numbers. Several examples are worked out in detail