Developing seasonal z-number regression for waste-disposal forecasting in a Taiwanese hospital

Abstract

This study develops a seasonal z-number regression (SZR) to forecast the daily generated amounts of clinical waste for recycling and related waste. The proposed SZR designs new z-number intervals based on least-squares support-vector regression and is combined with the seasonal-decomposition method. Z-numbers characterize the uncertainty of a variable using a fuzzy number and the associated reliability of this fuzzy number. Biogeography-based optimization is employed to select the parameters of the SZR model. In this study, SZR, long short-term memory, support-vector regression with a genetic algorithm, a generalized regression neural network, the autoregressive integrated moving average, and a recurrent neural network are applied to forecast the daily amounts of generated clinical waste for recycling and related waste. The empirical results indicate the following. First, the SZR model demonstrates better performance and robustness than the other approaches for the prediction of the daily amount of clinical waste for recycling and related waste. Second, the SZR model also exhibits superior performance compared to the other approaches with respect to different testing terms. The proposed SZR method can help experts in hospitals develop reasonable waste-disposal projects that can help waste-management systems effectively achieve operational reliability and economy.