veoexrmpzt
Dołączył: 21 Lut 2011
Posty: 371
Przeczytał: 0 tematów
Ostrzeżeń: 0/5
Skąd: England
|
 Wysłany: Wto 6:01, 05 Kwi 2011 Temat postu: mbt laarzen ace ups oeqb amc |
 |
|
Missing data based on time series analysis of tourist arrivals
Off cycle trend ,[link widoczny dla zalogowanych], dcx = (1 a Bn) cx = cx-cx (-12), remove the cyclical trends in the data shown in Figure 4 the data unit root test results are shown in Table I. The results show that the sequence is not the root of Is smooth . While observing the sequence dcx sample autocorrelation and partial correlation function , using least squares estimation model of dcx . The following model by AIC criterion to determine the order . Since the order of the season the greater the difference between the loss of more data , so there will be seasonal difference between the limit p = 2, q = 2 or less. In fact, through AIc minimum criteria, we found that within the season to establish AR (3) model, between season ARMA (4,[link widoczny dla zalogowanych],1) model , AIC value is minimal. But the difference order is too large, the loss of more data, not used. Thus the establishment Am ~ (2,1,2),[link widoczny dla zalogowanych], 2XAR/MA (3,0, o) model within the season, Ar (3), seasons ARMA (2,2) model is better. The AR (2) before the coefficient is not significant , the model graphs TECHNOLOGY INFORMATION SCIENCE & TECHNOLOGYINFORMATION t26 minor modifications , the parameter estimates and the statistics shown in Figure 5. 2.4 residual white noise test on the residuals test 10/10 = lO0% 95.4% residual white noise test passed. Finally the model is : (1-0.654283B.0.302522B) (1 + O.133462B +0.204937 B) 12log (x,[link widoczny dla zalogowanych],) = (1-0.344294B +0.784661 B) £ = 0.002906935 Method using composite predicted additional data . The modeling agreement with the above methods , in this not to say, only display the results. ARMA model in the establishment , the results are consistent with the above is still within the season selected AR (3), seasons ARMA (2,2) model is better. The eventual establishment of the model ( eg Equation 1 ). (1-0.606894B a 0.342234B) (1 + O.12729B +0.274467 B) Vlog (x): (1-0.458444B +0.732201 B )£,,= 0.002617039 (1) Comparative Analysis . Practical than the first method better. In summary,[link widoczny dla zalogowanych], the results of two complementary methods of data is quite consistent , the effect is quite good , can be used to practice foreign tourist arrivals forecast.
Post został pochwalony 0 razy
|
|
