Regression Modeling Selection Process-ECO6416

Regression Modeling Selection Process-ECO6416 - 7. F df1...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
Regression Modeling Selection Process When you have more than one regression equation based on data, to select the “best model", you should compare: 1. R-squares: That is, the percentage of variance [in fact, the sum of squares] in Y accounted for by variance in X captured by the model. 2. When you want to compare models of different sizes (different numbers of independent variables (p) and/or different sample sizes n), you must use the Adjusted R-Square, because the usual r-square tends to grow with the number of independent variables. r 2 a = 1 - (n - 1)(1 - r 2 )/(n - p - 1) 3. Standard deviation of error terms, i.e., observed y-value - predicted y-value for each x. 4. Trends in errors as a function of control variable x. Systematic trends are not uncommon. 5. The T-statistic of individual parameters. 6. The values of the parameters and its content to content underpinnings.
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 7. F df1 df2 value for overall assessment. Where df1 (numerator degrees of freedom) is the number of linearly independent predictors in the assumed model minus the number of linearly independent predictors in the restricted model; i.e., the number of linearly independent restrictions imposed on the assumed model, and df2 (denominator degrees of freedom) is the number of observations minus the number of linearly independent predictors in the assumed model. The observed F-statistic should exceed not merely the selected critical value of F-table, but at least four times the critical value. Finally in statistics for business, there exists an opinion that with more than 4 parameters, one can fit an elephant so that if one attempts to fit a regression function that depends on many parameters, the result should not be regarded as very reliable....
View Full Document

This note was uploaded on 10/06/2011 for the course ECO 6416 taught by Professor Staff during the Spring '08 term at University of Central Florida.

Ask a homework question - tutors are online