Skip to main content

Table 2 Univariate and multivariate ARIMA models of the weekly number of influenza cases and their performance, Abidjan, Cote d’Ivoire, 2007-2010

From: Effects of climatological parameters in modeling and forecasting seasonal influenza transmission in Abidjan, Cote d’Ivoire

Models

Performance criteria

AR (p)

MA (p)

Breakpoint (p)

Rainfall (RF)

Portmanteau test for residuals autocorrelation ‘white noise’

AIC

BIC

Coef.

Coef.

Coef.

Lag (EW −0 )

Lag (EW −5 )

Q (p)

Univariate Arima models

 ARIMA (0,0,2)

1305.828

1319.178

-

0.359 (0.000)

8.251 (0.000)

-

-

12.92 (0.227)

 ARIMA (0,0,3)

1303.898

1320.586

-

0.127 (0.017)

8.382 (0.000)

-

-

8.25 (0.604)

 ARIMA (0,0,5)

1301.275

1324.638

-

0.189 (0.001)

8.635 (0.000)

-

-

3.71 (0.959)

 ARIMA (1,0,0)

1306.692

1316.705

0.506 (0.000)

-

8.440 (0.000)

-

-

16.07 (0.097)

 ARIMA (1,0,1)

1298.622

1311.972

0.786 (0.000)

−0.374 (0.000)

9.326 (0.000)

-

-

2.377 (0.667)

 ARIMA (1,0,3)

1296.892

1316.917

0.989 (0.000)

−0.242 (0.000)

16.602 (0.000)

-

-

6.626 (0.760)

 ARIMA (2,0,0)

1296.889

1310.24

0.238 (0.000)

-

9.135 (0.000)

-

-

5.502 (0.855)

 ARIMA (2,0,5)

1302.240

1332.278

−0.526 (0.004)

0.156 (0.035)

8.837 (0.000)

-

-

1.851 (0.997)

Multivariate Arimax models

 ARIMAX (0,0,2) RF

1265.918

1285.797

-

0.332 (0.000)

6.467 (0.000)

0.02 (0.000)

0.023 (0.000)

13.08 (0.219)

 ARIMAX (0,0,3) RF

1263.395

1286.588

-

0.137 (0.016)

6.542 (0.000)

0.021 (0.000)

0.023 (0.000)

8.246 (0.604)

 ARIMAX (0,0,5) RF

1262.593

1292.412

-

0.149 (0.016)

6.905 (0.000)

0.018 (0.004)

0.025 (0.000)

4.359 (0.929)

 ARIMAX (1,0,0) RF

1263.068

1279.634

0.521 (0.000)

-

6.572 (0.000)

0.018 (0.001)

0.028 (0.000)

15.21 (0.124)

 ARIMAX (1,0,1) RF

1258.968

1278.847

0.736 (0.000)

−0.29 (0.002)

7.289 (0.000)

0.016 (0.007)

0.027 (0.000)

7.786 (0.649)

 ARIMAX (2,0,0) RF

1257.765

1277.644

0.192 (0.001)

-

7.258 (0.000)

0.017 (0.005)

0.026 (0.000)

5.894 (0.824)

  1. Note: AR autoregressive coefficient, MA moving average coefficient, Breakpoint coefficient of the breakpoint variable: a surge in the dynamic of the incidence of influenza cases from week 19 of 2009 following the occurrence of pandemic Influenza 2009, installation of new sentinel sites and the acquisition of new diagnostic technique by the national influenza laboratory (National Influenza Centre: NIC, Pasteur Institute), Q coefficient of the Portmanteau test for residuals autocorrelation, Coef. coefficient, p p-value, RF rainfall, ARIMA (X) introduction of independent variables in the model (multivariate Arima), AIC akaïke information criterion, BIC Bayesian information criterion, EW epidemiologic week, the subscripts indicate the number of weeks prior (0 being the same week, 1 the week prior, and so on)
Back to article page

BMC Public Health

ISSN: 1471-2458

Contact us