3D kernel regression (smoothing)

[isomitzi]

שלום לכולם

כמו שמציינת הכותרת, אני אשמח למעט עזרה בנושא:
my data set is ~800 cases of 2 predictors and 1 response
i would like to create (or fit?) a smooth s3D surface of the data and then use the smoothed surface, "slice" into 2D space and fit logistic regression to each slice
most people use 'ksmooth' but it does not apply to 3D space

any ideas?


תודה לכולם

ישראל