Table 3 Overall data structure for test period #2.

From: Forecasting the spread of COVID-19 based on policy, vaccination, and Omicron data

Country

Omicron variant usage

Vaccination coefficient

Prediction model

yijkl

DNK (l = 0)

Unused (\({c}_{k}\) = 0)

Unused (\({b}_{j,1}=\cdot \cdot \cdot ={b}_{j,4}=0\))

ARIMA (\({a}_{i,1}=\cdot \cdot \cdot ={a}_{i,5}=0\))

y0000

BiLSTM (\({a}_{i,1}=1\))

y1000

GAM (\({a}_{i,2}=1\))

y2000

lightGBM (\({a}_{i,3}=1\))

y3000

SEIR (\({a}_{i,4}=1\))

y4000

TSGLM (\({a}_{i,5}=1\))

y5000

1st-vaccination (\({b}_{j,1}\) = 1)

ARIMA (\({a}_{i,1}=\cdot \cdot \cdot ={a}_{i,5}=0\))

y0100

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

TSGLM (\({a}_{i,5}=1\))

y5100

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

1st, 2nd, 3rd vaccinations (\({b}_{j,4}\) = 1)

TSGLM (\({a}_{i,5}=1\))

y5400

Used (\({c}_{k}\) = 1)

Unused (\({b}_{j,1}=\cdot \cdot \cdot ={b}_{j,4}=0\))

ARIMA (\({a}_{i,1}=\cdot \cdot \cdot ={a}_{i,5}=0\))

y0010

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

1st, 2nd, 3rd vaccinations (\({b}_{j,4}\) = 1)

TSGLM (\({a}_{i,5}=1\))

y5410

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

\(\cdot \cdot \cdot\)

USA (l = 6)

Used (\({c}_{k}\) = 1)

1st, 2nd, 3rd vaccinations (\({b}_{j,4}\) = 1)

TSGLM (\({a}_{i,5}=1\))

y5416