Case Studies in Time Series II Periodic Behavior and Related issues

Case Studies in Time Series II Periodic Behavior and Related issues PowerPoint PPT Presentation


  • 109 Views
  • Uploaded on
  • Presentation posted in: General

Download Presentation

Case Studies in Time Series II Periodic Behavior and Related issues

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Presentation Transcript


2. An Example - Rollers

3. Rollers are to even out fabric coating Offset roller => Periodicity in thickness Period = Roller circumference Yt = thickness at time ARIMA model does not help!

4. Coating Thickness every ¼ inch

5. One “radian”

6. Fun with radians ? Half circle is p radians Assume all angles in [ - p, p ] Rotation slides along sine wave

7. Fun with trigonometry ? ? A Sin(wt+d) = A Sin(wt)Cos(d) + A Cos(wt)Sin(d) Got w ? Create Sin(wt), Cos(wt) [A Cos(d)] Sin(wt) + [A Sin(d)] Cos(wt) b1 Sin(wt) + b2 Cos(wt) Regression !!!

8. ? ? -- got w ! [A Cos(d)] Sin(wt) + [A Sin(d)] Cos(wt) b1 Sin(wt) + b2 Cos(wt) b12 + b22 = A2 [Cos2(d)+ Sin2(d)] = A2 b2 /b1 = Tan(d) d = Arctan(b2 /b1 ) From regression – recover Amplitude A Phase angle d Harmonics (wave form) 2w, 3w, 4w, ...

9. To support our students, we need a strong structural base, which offers outstanding opportunities. In the last two years, we have strengthened that base. On Centennial Campus, an innovative research park based on partnerships between the university and private companies or government entities, grew to include (*)72 partners employing 1300 people. Many students work at companies as interns. To support our students, we need a strong structural base, which offers outstanding opportunities. In the last two years, we have strengthened that base. On Centennial Campus, an innovative research park based on partnerships between the university and private companies or government entities, grew to include (*)72 partners employing 1300 people. Many students work at companies as interns.

10. X’X = (n/2)I (mostly) n/2(b12 + b22) = (n/2)A2 = regression SSq wj= 2pj/n = jth “Fourier Frequency” ( (n-1)/2 of these if n odd) PROC SPECTRA Calculate all Fourier Sin, Cos (n-1) Regress on all (multiple regression) orthogonal!!! Type I = Type III = “Periodogram Ordinate”

11. Polynomials Enough Xj terms (n-1) => fit data perfectly “Taylor Series Expansion” Goal: Enough, not too many Periodogram Enough terms (n-1)/2 sines, (n-1)/2 cosines => fit data perfectly! “Fourier Series Expansion” Goal: Enough, not too many

12. Periodogram Analysis Plot Periodogram ordinate I(wj) vs. wj Look for Periodicities Our case – can just do periodicities of rollers. Spectral density = smoothed, normalized periodogram

13.

14. A chancellor's team of faculty, students, trustees, community leaders, and campus designers produced a university master plan to define and upgrade campus neighborhoods over the next ten years. Planning and design for buildings funded by the $3.1 billion higher education bond began in 2000. The undergraduate science teaching lab is being designed as is the new engineering building, the CVM Building and renovations for Clark Hall in the first year of the buildout. A chancellor's team of faculty, students, trustees, community leaders, and campus designers produced a university master plan to define and upgrade campus neighborhoods over the next ten years. Planning and design for buildings funded by the $3.1 billion higher education bond began in 2000. The undergraduate science teaching lab is being designed as is the new engineering building, the CVM Building and renovations for Clark Hall in the first year of the buildout.

15. proc print data=outspec; where P_01 > 20 Obs FREQ PERIOD P_01 S_01 17 0.19635 32.0000 47.8072 0.92743 27 0.31907 19.6923 51.6533 1.17922 Two of our rollers are likely out of line. circumferences 5, 6, 8, 11, 16 and 20 inches One roller that is out of line would rotate once in 32 observations. With 4 observations per inch, that roller would have circumference 8 inches. A roller of circumference 19.69 quarter inches (4.9 inches) is not one of our known circumferences. Perhaps it is really the 5 inch one.

16. A chancellor's team of faculty, students, trustees, community leaders, and campus designers produced a university master plan to define and upgrade campus neighborhoods over the next ten years. Planning and design for buildings funded by the $3.1 billion higher education bond began in 2000. The undergraduate science teaching lab is being designed as is the new engineering building, the CVM Building and renovations for Clark Hall in the first year of the buildout. A chancellor's team of faculty, students, trustees, community leaders, and campus designers produced a university master plan to define and upgrade campus neighborhoods over the next ten years. Planning and design for buildings funded by the $3.1 billion higher education bond began in 2000. The undergraduate science teaching lab is being designed as is the new engineering building, the CVM Building and renovations for Clark Hall in the first year of the buildout.

17. data waves; pi=4*atan(1); set fabric; s5 = sin(2*pi*location/5); c5 = cos(2*pi*location/5); s6 = sin(2*pi*location/6); c6 = cos(2*pi*location/6); s8 = sin(2*pi*location/8); c8 = cos(2*pi*location/8); s11 = sin(2*pi*location/11); c11 = cos(2*pi*location/11); s16 = sin(2*pi*location/16); c16 = cos(2*pi*location/16); s20 = sin(2*pi*location/20); c20 = cos(2*pi*location/20); proc reg; model Y = s5--c20/ss1 ss2; T_5: test s5=0, c5=0; T_6: test s6=0, c6=0; T_8: test s8=0, c8=0; T_11: test s11=0, c11=0; T_16: test s16=0, c16=0; T_20: test s20=0, c20=0;

18. The TEST statement results are 5 inch 6 inch 8 inch 37.0 0.22 20.08 Pr > F = <.0001 .8018 <.0001 11 inch 16 inch 20 inch 3.27 2.72 1.36 Pr > F = .0387 .0670 .2586

19. Theoretical Spectral Density ( g(h) = cov{Yt, Yt-h} )

20. Generate Theoretical Densities Data AR_1; do t = 1 to 512; e= normal(123); Y = .8*Y + 2*e ; if Y=. then Y=0; X = -.8*X + 2*e; if X=. then X=0; drop e; output; end; proc spectra P S adjmean out=arspec; Var Y X; weight 1 2 3 4 5 6 7 6 5 4 3 2 1; data arspec; set arspec; Ytheory = 4/(2*3.14159*(1.64 - 1.6*cos(freq))); Xtheory = 4/(2*3.14159*(1.64 + 1.6*cos(freq))); label S_01 = "Y (.8)"; label S_02 = "X (-.8)"; proc gplot; plot (S_01 Ytheory)*freq/overlay; symbol1 v=dot i=none c=red h=.6; symbol2 v=none i=join c=black w=3; plot (S_02 Xtheory)*freq/overlay; run;

21. Overlay Theoretical & EstimatedDensities

22. Plant Enzyme Activity

23. proc spectra whitetest; The SPECTRA Procedure Test for White Noise for Variable Y M-1 14 Max(P(*)) 11932.89 Sum(P(*)) 34590.36 Fisher's Kappa: (M-1)*Max(P(*))/Sum(P(*)) Kappa 4.829682 Bartlett's Kolmogorov-Smirnov Statistic: Maximum absolute difference of the standardized partial sums of the periodogram and the CDF of a uniform(0,1) random variable. Test Statistic 0.255984 Approximate P-Value 0.3180

24. Lesson learned F test was significant (regression) Spectral tests not significant. Price paid (power) for not knowing frequency !! ? Researcher expected this frequency in advance.

25. ln {NC Retail Sales (million $)}

26. Residual Periodogram & PROC GAM fit

27. Transfer Functions X-> Y Yt = __ + __Xt + __Xt-1 +__Xt-2 + … + et Recall: Xt has Fourier Expansion Yt has Fourier Expansion X sinusoid -> Y sinusoid at same frequency Amplitude (A) may change Phase shift (d)

28. Neuse River flows – up & down stream

29. Gain = amplitude change

30. Thanks !

  • Login