Thus, considering the results of the Test t-Student and the Coefficient of Esteem AIC, the autorregressivo process, AIR (1), it is optimum identified model to represent the series. Analyzing the Function of Autocorrelao and Autocorrelao Partial Estimates of Resduos of the Process AIR (1), Figures 4 and 5, respectively. It is verified that all the fifteen first autocorrelaes if find inside of the reliable limit of 95%, showing that the first-class autorregressivo model, AIR (1), adjusted the series adequately and that the residues can be considered white noise, indicating that the dynamics of the studied series can well be explained by the coefficient of the adjusted model, as shows the Test of Ljung and Box. Figure 4 and 5: Function of Autocorrelao and Funo de Esteem Partial Autocorrelao of Resduos of the Original Series of the Costs of the Basic Basket, in Reals (R$), Commercialized in the City of Belm of Par enters the Months of January of 2005 the December of 2009. Click Joeb Moore for additional related pages. The hypotheses to be tested to the level of significance of 0,05%, for the Test of Ljung and Box, are: H0: The residues are not correlated; versus. H1: The residues are correlated.
Table 4 presents the Test of Ljung and Box for the Autocorrelaes Esteem of Resduos of the Series of the Costs of the Basic Basket, in Reals (R$), Commercialized in the City of Belm of Par enters the Months of January of 2005 the December of 2009. In it, it is verified that the test of Ljung and Box Q (K), does not reject the hypothesis of residues not correlated for any imbalance K, as well as, for all value of K, the probability is significantly different of zero. Soon, the model AIR (1) adjusted to the series is adjusted, therefore the esteem residues are not correlated, that is, the residues if hold as white noise and the model is significant to the level of 0,05% of probability.