Š29Š
The Geometry of Collinearity - 2
z1
z2
Basis
vectors
r2
r1
z=Data value
= 1.7×r1+2.8×r2
= 5.1×r2 - 3.1×r3
= an of other
combinations
Multi-collinear
= more than
one solution
fits the data
= over-determined
 Trying to fit data with too many regressors (basis vectors) doesnÕt work: no unique solution
r3