APPLICATION OF TRAPEZOIDAL FUZZIFICATION APPROACH (TFA) AND PARTICLE SWARM OPTIMIZATION (PSO) IN FUZZY TIME SERIES (FTS) FORECASTING

摘要

The trapezoidal fuzzification approach (TFA) model implemented in this paper utilizes aggregation and particle swarm optimization (PSO) to reduce the mismatch between the forecasted data and the actual data using maximum temperature data of Zaria for the period 1990-2003, obtained from Nigerian Meteorological Agency (NIMET) Zaria. The defuzzification module (including the developed PSO algorithm using C-#) of this model is then implemented on the hitherto fuzzified maximum temperature data so as to obtain forecasts. Statistical measures of MSE and MAPE are used to test the reliability of the model.