ABSTRACT
In Chapter I it was shown by example how the variability of least square estimators can be seriously inflated for collinear data. It was also shown how the ridge estimator of Hoerl and Kennard can be used to reduce this variability. The ridge estimator was originally proposed by Hoerl (1962) because data in chemical processes was often multicollinear. Since then it has found applications in a wide variety of diverse fields.
