Statistical Summary

The statistics of Ljung-Box, it is a statistical test of null hypothesis of autocorrelao absence. Thus, if for all imbalance the statistics, we can consider that the residues if hold as white noise and that, therefore, the adjusted model is adjusted. In case that the model is not adjusted, the cycle is repeated, turning it the identification phase. A procedure very used is to identify not alone an only model, but some models that then will be esteem and verified. When if it gets a satisfactory model, is transferred to the last stage of the methodology of Box-Jenkins, that constitutes the main objective of the methodology: to carry through forecasts. 2.2.2.Critrio of Validation According to Russian (2002), one of the criteria very used to choose optimum model is the criterion of Percentile Error Absolute Medium of Previso (MAPE). The MAPE is calculated from forecasts a step to the front generated for each esteem model.

Where it is the observed value of the series, foreseen value it is the number of forecasts. 3. Anlise and Discurso of the Data Table 1 presents the Statistical Summary of the Costs of the Basic Basket, in Reals (R$), Commercialized in the City of Belm of Par of January of 2005 the December of 2009. In it, it is verified that throughout this period, the monthly average cost of the basic basket was of R$ 174,66 with variability around the average cost of R$ 23,49. Table 1: Statistical summary of the Costs of the Basic Basket, in Reals (R$), Commercialized in the City of Belm of Par of January of 2005 the December of 2009. Measures Average value 174,66 Medium 165,14 Shunting line Standard 23,49 as the average cost is greater that the medium cost, can be said that the series presents positive asymmetry.

Comments are closed.