ABSTRACT
This chapter shows that ridge regression acts by scaling the singular values of X-X in such a way that the smaller singular values are increased in order to stabilize the solution to the normal equations. It looks at a dataset from the NYC Taxi and Limousine Commission. The chapter explains some of spatial data into our predictive model, and explore the performance of ridge regression in this context. Ridge regression maintains the rotational invariance of ordinary least squares. Ridge regression can be understood as a Bayesian estimator to the prior distribution and likelihood functions implied by assigning independent, identically distributed, normally random variables to the components of β and setting the likelihood function to that of the standard regression model. The chapter deals with pairs of neighborhoods corresponding to where each cab picked up a passenger and then the neighborhood where they dropped the passenger off.
