Spatiotemporal Study of COVID-19 in Fars Province, Iran, October-November 2020: Establishment of Early Warning System
Table 5
The frequency 55 369 forecasted cases during 1 October–14 November 2020 accompanied by the results of time series analysis in the 26 cities of Fars province, Iran.
Time series analysis results
City
Forecasted cases
Model
Residual analysis tests ()
Order of min AICC YW
AIC
BIC
n
%
Ljung-box
Diff sign points
Rank test statistic
McLeod-Li
Turning points
Jarque-bera
Shiraz
14895
26.9
MA (8)
0.230
0.30
0.69
0.030
0.100
<0.001
0
8280
8303
Lar
2762
4.99
SARMA12 (2, 2)
0.400
0.9
0.49
0.040
0.560
<0.001
0
1098
1101
Jahrom
1716
3.10
SARMA12 (3, 4)
0.770
0.73
0.37
<0.001
0.010
<0.001
0
1221
1225
Fasa
1772
3.20
ARMA (1, 1)
0.460
0.73
0.12
<0.001
0.090
<0.001
0
1371
1371
Kazerun
1661
3
AR (1)
0.440
0.30
0.04
0.020
0.220
<0.001
0
1267
1266
Marvdasht
2215
4
SARMA12 (1, 1)
0.610
0.9
0.04
0.010
0.560
<0.001
0
1247
1247
Darab
1440
2.60
ARMA (1, 1)
0.960
0.02
0.80
0.040
0.180
<0.001
0
1360
1361
Abade
1661
3
SARMA12 (1, 1)
0.960
0.42
0.60
0.040
0.180
<0.001
0
1360
1361
Zarindasht
1883
3.40
AR (1)
0.860
0.03
0.85
0.010
0.950
0.030
0
1357
1356
Firoozabad
1495
2.70
SARMA12 (1, 2)
0.060
0.08
0.21
0.030
0.010
<0.001
0
1391
1389
Lamerd
1440
2.60
SARMA12 (2, 2)
0.190
0.21
0.34
<0.001
0.0002
0.030
0
1364
1368
Mamasani
1495
2.70
SARMA12 (1, 2)
0.580
0.04
0.32
0.040
0.490
<0.001
0
1381
1380
Ghirkarzin
1440
2.60
ARIMA (3, 1)
0.980
0.73
0.51
0.020
0.450
0.010
0
1364
1366
Safashahr
1440
2.60
ARIMA (1, 1)
0.500
0.45
0.76
0.0010
0.350
0.020
0
1348
1348
Sepidan
1993
3.60
SARMA12 (3, 1)
0.130
0.30
0.92
0.030
0.950
0.040
0
1344
1342
Stahban
1440
2.60
ARMA (1, 1)
0.980
0.2
0.81
<0.001
0.180
<0.001
0
1391
1392
Eghlid
1440
2.60
MA (1)
0.140
0.73
0.17
0.020
0.670
<0.001
0
1391
1388
Farashband
1440
2.60
SARMA12 (2, 2)
0.920
0.73
0.40
<0.001
0.040
0.001
0
1369
1370
Pasargad
1661
3
MA (1)
0.060
0.04
0.47
0.010
0.490
0.050
0
1334
1332
Neireez
1440
2.60
AR (1)
0.230
0.45
0.34
0.002
0.450
0.010
0
1364
1363
Arsenjan
1440
2.60
ARMA (2, 1)
0.080
0.42
0.14
0.030
0.830
<0.001
0
1372
1367
Rostam
1440
2.60
AR (1)
0.090
0.07
0.20
<0.001
<0.001
<0.001
0
1321
1321
Khonj
1440
2.60
SARMA12 (2, 1)
0.680
0.06
0.05
<0.001
0.830
<0.001
0
1295
1291
Kharame
1440
2.60
SARMA12 (1, 3)
0.450
0.43
0.23
0.001
0.001
0.010
0
1356
1357
Bavanat
1440
2.60
MA (1)
0.220
0.91
0.02
<0.001
0.310
<0.001
0
1308
1306
Sarvestan
1440
2.60
SARMA12 (2, 2)
0.540
0.9
0.96
0.010
0.590
0.130
0
1260
1258
AR, autoregressive; MA, moving average; ARMA, autoregressive moving average; ARIMA, autoregressive integrated moving average; SARMA, seasonal autoregressive moving average. It is noteworthy to mention that if four out of the six residual analysis test statistics are statistically significant, it is enough to say that the model is a good fit; based on residual analysis McLeod- Li test statistic needs to be significant and the other five tests should be greater than or equal to 0.05; in addition, the order of min AICCYW, which assesses the mean of white noise residual, needs to be zero. The lower the AIC/BIC scores, the better the model fits.