Table 4 Comparison of BeCaked and top-tier forecasting-specialized methods at country level with 1-step forecast.

From: BeCaked: An Explainable Artificial Intelligence Model for COVID-19 Forecasting

 

BeCaked

ARIMA

Ridge

LASSO

SVR

DTR

RFR

GBR

Australia

\(R^2\)

0.99112

0.98329

0.99337

\(\mathbf {0.99531}\)

\(-0.48798\)

0.98342

0.96787

0.98299

MAPE

0.04703

0.10236

\(\mathbf {0.01093}\)

0, 04078

0.85609

0.10204

0.14942

0.10344

Italy

\(R^2\)

\(\mathbf {0.96381}\)

0.94660

0.92646

0.88220

\(-22.56497\)

0.94889

0.90082

0.94876

MAPE

0.00496

0.01655

\(\mathbf {0.00024}\)

0.05999

0.93214

0.01684

0.02570

0.02165

Russia

\(R^2\)

0.92922

0.91663

\(\mathbf {0.98479}\)

0.92367

\(-108.89720\)

0.91643

0.87353

0.91854

MAPE

\(\mathbf {0.00004}\)

0.02856

0, 00007

0.00938

0.77448

0.02851

0.04442

0.02938

Spain

\(R^2\)

\(\mathbf {0.98765}\)

0.97214

0.98591

0.98362

\(-10.84390\)

0.97212

0.94968

0.97053

MAPE

\(\mathbf {0.00174}\)

0.01418

0.00325

0.01881

0.74045

0.01406

0.02173

0.01522

United Kingdom

\(R^2\)

\(\mathbf {0.99833}\)

0.98713

0.99764

0.96375

\(-449.89408\)

0.98748

0.96991

0.98463

MAPE

0.00030

0.00255

\(\mathbf {0.00014}\)

0.00437

0.52586

0.00249

0.00397

0.00282

United States

\(R^2\)

\(\mathbf {0.99688}\)

0.98676

0.99648

0.99461

\(-9.99942\)

0.98676

0.96834

0.98633

MAPE

0.00552

0.02644

\(\mathbf {0.00004}\)

0.00069

0.62264

0.02642

0.04185

0.02687

  1. Best results are in [bold].