A PRACTICAL TECHNIQUE FOR ESTIMATING GENERAL REGRESSION SURFACES
LOCKHEED MISSILES AND SPACE CO INC PALO ALTO CA PALO ALTO RESEARCH LAB
Pagination or Media Count:
Let X be a p component random vector variable with transpose X is identically equal to X sub 1,..., X sub j,..., X sub p and Y be a one- dimensional random variable with a joint continuous distribution of density fx, y. The regression of Y given X x is EY such that X x the integral from minus infinity to plus infinity of yfx,ydy divided by the integral from minus infinity to plus infinity of fx,ydy. Using consistent nonparametric estimators of the densities, EY such that X x can be approximated by a ratio of two general polynomials where the coefficients of the polynomials are computed as a function of the observed sample.
- Statistics and Probability