evaluating loess fit

Hello Everyone,

I am trying to evaluate the fits of data using loess() in R. I cannot seem to find any established methods for determining the quality of a loess fit. I have been using the median absolute deviation of the residuals as an estimate of the "goodness of fit" and the amount of variability in the response. This is something like a robust RMSE estimate.

Does anyone have a suggestion for additional methods. I am mainly interested in the variance of the data about the fit, not so much the fit itself.