The ENSO data is analyzed in an example in the PROC TRANSREG documentation. (ENSO is an acronym for El Niño-Southern Oscillation.) To illustrate fitting a smooth curve to a scatter plot, I'll use the SasHelp.enso data set, whichĬontains data on the "southern oscillation," a cyclical phenomenon in atmospheric pressure that is linked to the El Niño and La Niña temperature oscillations in the Pacific Ocean. These criteria try to produce a curve that fits the data well, but does not interpolate or overfit the data. The TRANSREG procedure supports the corrected Akaike criterion (AICC), the generalized cross-validation criterion (GCV), and Schwarz’s Bayesian criterion (SBC), among others. The TRANSREG procedure enables you to specify many options that control the fit, including the criterion used to select the optimal smoothing parameter. You can fit penalized B-splines to data by using the TRANSREG procedure or by using the PBSPLINE statement in the SGPLOT procedure. This article describes penalized B-splines loess curves, and thin-plate splines.
#Sas fitplot software#
In contrast, SAS software supports several other methods for which the software can automatically choose a smoothing parameter that maximizes the goodness of fit while avoiding overfitting. This means that you can specify a smoothing spline parameter that appears to fit the data, but the selection is not based on an objective criterion. I ended the post with a cautionary note:įrom a statistical point of view, the smoothing spline is less than ideal because the smoothing parameter must be chosen manually by the user. My last blog post described three ways to add a smoothing spline to a scatter plot in SAS.
0 kommentar(er)
