MGEB11H3 Chapter Notes - Chapter 16.1-16.6: General Linear Model, F-Test, Multicollinearity
9 Apr 2016
School
Department
Course
Professor
Document Summary
General linear model: muliple regression analysis linear refers to fact b0, b1, ,bp have exponents 1. Does not imply relaionship b/w x and y is linear. Recall: simple case w/ 1 variable x to esimate y (uses straight line relaionship_ o. Therefore, z1 = x1 y = b0 + b1x1 + e simple irst order model w/ 1 predictor var. Ex: reynolds inc. , manufacturer or industrial scales and laboratory equip, invesigate relaionship b/w length of employment, # of electronic laboratory sold. Est reg = ^y = 111 + 2. 38x. Although data shows linear relaionship is signiicant and has r2 = 78. 1% (explains high variability) 2 + e second order model w/ 1 predictor var. New est reg: ^y = 45. 3 + 6. 34x1 0. 0345x1. At a=0. 05, overall model = signiicant (meaning, adding x1. R2 a = 88. 6% increased it provided by esimate. Interacion = 2nd order model w/ 2 predictor variables (observaions for y + 2 independent var x1 and x2)
