analysis of fit to cosine waveI am planning to analyze circadian rhythm data and can fit it to the Savitzky-Golan smoother with a cubic degree and local cosine weighting. This provides a visual with a confidence of fit. However, is there a way to statistically determine the quality of the fit to the cosine waveform? I have tried saving the smoother formula to the data table then launching the analysis using the upper and lower smoother values in a Sheffe Cubic macro with the x-time covariant but I am not sure if this is the best way to analyze fit. Any advice?
Thanks,
Arthur DeCarlo
Advanced Statistical Modeling
