14.03.2014 Views

Modeling and Multivariate Methods - SAS

Modeling and Multivariate Methods - SAS

Modeling and Multivariate Methods - SAS

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

358 Performing Time Series Analysis Chapter 14<br />

Time Series Comm<strong>and</strong>s<br />

N 2 2<br />

where f i<br />

= i⁄<br />

N are combined to form the periodogram If ( i<br />

) = --- ( a , which represents the intensity<br />

at frequency f i .<br />

2 i<br />

+ b i<br />

)<br />

The periodogram is smoothed <strong>and</strong> scaled by 1/(4π) to form the spectral density.<br />

The Fisher’s Kappa statistic tests the null hypothesis that the values in the series are drawn from a normal<br />

distribution with variance 1 against the alternative hypothesis that the series has some periodic component.<br />

Kappa is the ratio of the maximum value of the periodogram, I(f i ), <strong>and</strong> its average value. The probability of<br />

observing a larger Kappa if the null hypothesis is true is given by<br />

q<br />

Pr( k > κ) = 1–<br />

(–<br />

1) j q<br />

<br />

<br />

max jk 1 – --- , 0 <br />

q – 1<br />

j <br />

q <br />

j = 0<br />

where q = N /2 if N is even, q =(N -1)/2 if N is odd, <strong>and</strong> κ is the observed value of Kappa. The null<br />

hypothesis is rejected if this probability is less than the significance level α.<br />

For q - 1 > 100, Bartlett’s Kolmogorov-Smirnov compares the normalized cumulative periodogram to the<br />

cumulative distribution function of the uniform distribution on the interval (0, 1). The test statistic equals<br />

the maximum absolute difference of the cumulative periodogram <strong>and</strong> the uniform CDF. If it exceeds<br />

a⁄<br />

( q)<br />

, then reject the hypothesis that the series comes from a normal distribution. The values a = 1.36<br />

<strong>and</strong> a = 1.63 correspond to significance levels 5% <strong>and</strong> 1% respectively.<br />

Figure 14.5 Spectral Density Plots<br />

Save Spectral Density<br />

Save Spectral Density creates a new table containing the spectral density <strong>and</strong> periodogram where the<br />

(i+1)th row corresponds to the frequency f i =i/N (that is, the ith harmonic of 1 / N).<br />

The new data table has these columns:<br />

Period is the period of the ith harmonic, 1 / f i .<br />

Frequency is the frequency of the harmonic, f i .

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!