QTM 100 Lecture Notes - Lecture 21: Kelvin Mercer, Regression Analysis, Dependent And Independent Variables

• School: Emory University
• Department: Quantitative Theory & Methods
• Course: QTM 100
• Professor: Jacobson Jeremy

Document Summary

Have many variables, but some are correlated to each other. How to narrow down to the core variables to explain as much as you can in the simplest way possible. Started from fact that in the background that all definitions are based on. Sums of squares total = sums of squares regression + sums of squares error. R2 = ssreg/sstotal = (sstotal - sserror)/ ss total. Take ss reg and divide by ss total. Measure of how well the model (line or plane) explains the change in y values of your response variable points. |correlation| = the square root of r2 (to up sign ( )) The square root gives the absolute value of the correlation. Paperback(*1) 187 g less than hardcover, plug 1 to account for the 187 (26) If two variables are correlated, you only need one because of their relation. In experiments, we can try to control this.