This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Box Jenkins Method Figure (1) Since the Figure (1) shows that the Autocorrelation of original Turbidity decays very slowly at the Nonseasonal level, it is Nonstationary . And the Autocorrelation has spikes at lags 2, 4, 6.. The numbers of lags are multiplied by 2, so I should use Second difference in Turbidity. Figure (2) The Autocorrelation and Partial Autocorrelation of Second differences of the Turbidity are given in Figure (2) the Autocorrelation dies down quickly, it is Stationary now, the second difference gives Stationary , but the Standard Deviation is too large, it is 365.45. Figure (3a) From Figure (3a) , after using Predifferencing Transformation, put the Log in the variable Turbidity, the standard deviation is much smaller, it is 1.007536. Figure (3b) H0: p1=p2=p3=p4=p5=p6=0 H1: at least one is nonzero X = 111.87 and pvalue of less than 0.0001, so reject Ho From Figure(3b) , since the Logarithm before second difference of capacity utilization is Not white noise . So we proceed to the modeling and estimation stage. Therefore, we should use logarithm before Second difference. Figure (3c) From the Logarithm before second difference , While the spikes at lags 2, 4, 6, 8, 10 and 18 in the Partial Autocorrelation of Second differences of the Turbidity and they die down fairly quickly. Partial Autocorrelation of Second differences of the Turbidity and they die down fairly quickly....
View
Full
Document
This note was uploaded on 01/12/2011 for the course STAT 646 taught by Professor J during the Fall '10 term at NJIT.
 Fall '10
 j
 Correlation

Click to edit the document details