Random Fuzzy Variable based Uncertainty Modelling for the Prediction of Human Development Index using CO2 Emission Data

The instruments used for the measurements are often accredited with some amount of error or uncertainty. These ambiguities and uncertainties may affect the process of decision making, quality assessment, risk analysis etc. Although, there are several high-end devices which can comfortably calculate the uncertainty in the measurement, the expression and representations of such uncertainty is a major bottleneck. From several years, Random Fuzzy Variables (RFVs) have provided the solution to this problem of uncertainty modelling. In this paper, we have modelled the uncertain readings of CO2 emissions from the satellites and earthly devices using the RFVs. Later, the Human Development index (HDI) is predicted using the above modelled data. Three most widely used machine learning (ML) approaches are utilized for the prediction purpose. These approaches are: kernel extreme learning machines (KELM), generalized regression neural network (GRNN), and support vector regression (SVR). Experimental results have shown the efficacy of the proposed approach with better RMSE in HDI predictions as compared to the traditional approaches.