NOTE ON A NEW COMPUTATIONAL DATA SMOOTHING PROCEDURE SUGGESTED BY MINIMUM MEAN SQUARE ERROR ESTIMATION
Abstract
The form taken by minimum mean square error estimation procedures is investigated for the case where the dependence of the observed data on the parameters can, with good approximation, be considered to contain only constant, linear, and quadratic terms. The resulting procedures can be considered as alternatives to the iterative procedures used in least squares.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1965
- Accession Number
- AD0615935
Entities
People
- P. Swerling
Organizations
- RAND Corporation