Multiple linear regression in databases
Sophie Leuenberger · 2014
Multidimensional statistical models such as multiple linear regression models are usually computed outside a data base management system (DBMS). In this report, we study how a multiple linear regression model can be computed inside a DBMS. The concept of “summary matrices”, which was introduced by Ordonez[5] as a tool to compute statistical models inside a DBMS, is presented and adapted for the case of multiple linear regression. We will consider two different approaches of implementation, study the performance of both of the alternatives, and figure out which one is better suited in which situation.