Development Of An Influenza Outbreak Forecasting Model Using Time Series Analysis Methods.

First, concentrating on the seasonal (s = 52) lags, the ACF shows a strong peak at h = 1s and lesser, but still statistically significant peaks at h = 3s , 4s, 5s, and 7s. The PACF shows statistically significant peaks at only h = 1s. It appears that: • the ACF is cutting off after lag 1s and the PACF is tailing off in the seasonal lags • the ACF is cutting off after lags 3s, 4s, 5s, or 7s, and the PACF is tailing off in the seasonal lags, or • the ACF and PACF are both tailing off in the seasonal lags, Using Table 3.3 from Shumway11, this suggests either • an SMA of order Q = 1, (0,1,1)52 • an SMA of order Q = 3, (0,1,3)52 • an SMA of order Q = 4, (0,1,4)52 • an SMA of order Q = 5, (0,1,5)52 • an SMA of order Q = 7, (0,1,7)52, or • an SARMA of orders P = 1 (because of the spike at 1s) and Q = 1, (1,1,1)52