Table 4 Comparison of computational time and prediction accuracy for different data-driven forecasting methods applied to the susceptible population S(t) of the proposed SIRD model.

From: Analysis, control, and forecasting the dynamics of SIRD models with saturated treatment and nonlinear incidence

Method

Time (s)

Train RMSE

Test RMSE

LMRC

3.06927

\(2.17528 \times 10^{-4}\)

\(1.87363 \times 10^{-4}\)

ESN

0.376473

\(1.89397 \times 10^{-4}\)

\(4.32770 \times 10^{-3}\)

GRU

9.37012

\(9.20545 \times 10^{-3}\)

\(6.95501 \times 10^{-2}\)

LSTM

10.8283

\(6.31987 \times 10^{-3}\)

\(6.07623 \times 10^{-2}\)

Back to article page