/* This is monthly influenza data taken from Robert H. Shumway, APPLIED STATISTICAL TIME SERIES ANALYSIS, 1988, p. 194. The table is labeled "Country X Monthly Cases of Influenza for 10 Years (1965 - 1974)." */ options pagesize=60 linesize=74 nodate; goptions device=win gsfname=plot rotate=landscape gsfmode=append target=hpljs3; *border; data influ; format date monyy5.; input date:monyy5. flu @@; cards; jan65 633 feb65 488 mar65 852 apr65 1097 may65 2447 jun65 8111 jul65 16483 aug65 25156 sep65 11980 oct65 7294 nov65 1284 dec65 4674 jan66 3084 feb66 878 mar66 1267 apr66 2530 may66 3175 jun66 3709 jul66 5180 aug66 14263 sep66 10244 oct66 13688 nov66 10277 dec66 6370 jan67 1718 feb67 1474 mar67 2277 apr67 4535 may67 7467 jun67 10791 jul67 7217 aug67 21100 sep67 14033 oct67 21953 nov67 9706 dec67 2782 jan68 2818 feb68 1150 mar68 2276 apr68 4388 may68 27559 jun68 50000 jul68 27650 aug68 31580 sep68 13729 oct68 11441 nov68 3410 dec68 1336 jan69 988 feb69 1055 mar69 2467 apr69 4867 may69 8718 jun69 38448 jul69 35590 aug69 32118 sep69 18391 oct69 11253 nov69 4912 dec69 2190 jan70 1287 feb70 1106 mar70 2092 apr70 5019 may70 7269 jun70 11145 jul70 25515 aug70 25434 sep70 24076 oct70 9605 nov70 5566 dec70 3765 jan71 1697 feb71 1666 mar71 4097 apr71 5808 may71 8963 jun71 17291 jul71 15885 aug71 14330 sep71 14522 oct71 8184 nov71 5956 dec71 4454 jan72 2005 feb72 1990 mar72 3021 apr72 4950 may72 8227 jun72 8251 jul72 15161 aug72 20986 sep72 11168 oct72 5794 nov72 4310 dec72 4102 jan73 2377 feb73 2240 mar73 4245 apr73 6452 may73 15736 jun73 25880 jul73 21354 aug73 26784 sep73 12722 oct73 12353 nov73 7182 dec73 11419 jan74 2563 feb74 2691 mar74 3853 apr74 6326 may74 9362 jun74 10685 jul74 13148 aug74 39093 sep74 19720 oct74 16492 nov74 7035 dec74 12224 ; title 'Influenza in Country X by Month Jan. 1965 - Dec. 1974'; axis1 label=('Year'); axis2 order=(0 to 60000 by 5000) label=(angle=90 'Cases of Influenza'); proc gplot data=influ; plot flu*date / haxis=axis1 vaxis=axis2; symbol1 i=join; format date year4.; run; /* Here we use the autocorrelation function to try to uncover signs of seasonality in flat (non-trending) data. Notice that we do not difference the data before examining the autocorrelation functions for the spikes at the seasonal lags j = 12, 24, 36, and 48 because the data does not have any slope to it (i.e. it is flat). */ proc arima data = influ; identify var=flu nlag=48; run;