Least Absolute Values(LAV) Regression.

LAV regression, also known as L-1 norm and least absolute deviations, is highly resistant to outliers but requires extensive computations. New algorithms for LAV regression based on linear programming methods were developed by Armstrong and Frome (1976a,1976b,1976c,and 1979). The first application of LAVs and related resistant regression methods to a nonlinear regression function was presented for a model that is widely used in pharmacokinetics (Frome and Yakatan, 1980).

Recent Applications of LAV Regression:

Back to Ed Frome's Home Page

Last Modified 10Jan 2003FromeEL@ORNL.gov (touches: [an error occurred while processing this directive] )