Nations performance evaluation during SARS-CoV-2 outbreak handling via data envelopment analysis and machine learning methods

Abstract

Nowadays, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) with COVID-19 public name (short for coronavirus disease 2019) encompassed the world and nations are trying to manage crisis with maximum medicine capacity, but how successful are they? Data envelopment analysis (DEA) is a powerful tool that answers the question with measuring the efficiency of nations, but the challenge is here that the COVID-19 data grow and change rapidly. So, efficiency measurement is difficult at any time of the epidemic because needs to re-implementing DEA models. At this point, Machine Learning (ML) comes to help, in way that efficiency scores prediction is feasible with supervised learning on DEA results, but accurate prediction in small-scale data is the next challenge. This paper investigates integrated DEA and ML (DEAML) to fix challenges. First, a relational two-stage model with desirable-undesirable variables is proposed to measure the efficiency of 50 nations by 5 December 2020. Then, a multi-layer perceptron (MLP) network with a Limited memory BFGS (L-BFGS) optimisation algorithm is proposed to predict the efficiency of nations at any time of the epidemic. The results are analysed and discussed. ABBREVIATIONS: CCR: Charnes–Cooper-Rhodes; BPNN: Back-Propagation Neural Network; GANN: Genetic Algorithm integrated with Neural Network; SVM: Support Vector Machines; ISVM: Improved Support Vector Machines; OECD: Organization for Economic Co-operation and Development; BCC: Banker-Charnes–Cooper; CART: Classification And Regression Trees; BT: Boosted Tree; PCR: Polymerase Chain Reaction; ICU: Intensive Care Unit; VRS: Variable Return to Scale; MSE: Mean Squared Error; GDP: Gross Domestic Product