L1-norm-robust-regression

This software provides a Fortran95 implemented routine to call and obtain the L1-norm regression coefficients.

The "standard" linear regression minimizes the squared error. (L2-norm-regression). It allows for a simple solution process, hence its popularity.
But it is not robust to outlier points.
L1-norm regression is robust with respect to outliers but the solution algorithm is more difficult. It minimizes the sum of absolute deviations.