On the prediction of Covid-19 time series: an intuitionistic fuzzy logic approach

Abstract

This paper presents a time series analysis of a novel coronavirus, COVID-19, discovered in China in December 2019 using intuitionistic fuzzy logic system with neural network learning capability. Fuzzy logic systems are known to be universal approximation tools that can estimate a nonlinear function as closely as possible to the actual values. The main idea in this study is to use intuitionistic fuzzy logic system that enables hesitation and has membership and non-membership functions that are optimized to predict COVID-19 outbreak cases. Intuitionistic fuzzy logic systems are known to provide good results with improved prediction accuracy and are excellent tools for uncertainty modelling. The hesitation-enabled fuzzy logic system is evaluated using COVID-19 pandemic cases for Nigeria, being part of the COVID-19 data for African countries obtained from Kaggle data repository. The hesitation-enabled fuzzy logic model is compared with the classical fuzzy logic system and artificial neural network and shown to offer improved performance in terms of root mean squared error, mean absolute error and mean absolute percentage error. Intuitionistic fuzzy logic system however incurs a setback in terms of the high computing time compared to the classical fuzzy logic system.