Re:

[Mathan]

Good.. Beta weights is still unanswered..

On Mon, Jun 7, 2010 at 1:47 PM, Allabux Jaffer <bux...@gmail.com> wrote:

---------- Forwarded message ----------
From: Jaffer, Allabux <Allabux...@nielsen.com>
Date: Mon, Jun 7, 2010 at 1:45 PM
Subject:
To: bux...@gmail.com

 

 

 

 

R² (R-sq)

Coefficient of determination; indicates how much variation in the response is explained by the model. The higher the R² , the better the model fits your data. The formula is:

Notation

	=	ith observed response value
	=	mean response
	=	ith fitted response

Adjusted R²

Accounts for the number of predictors in your model and is useful for comparing models with different numbers of predictors. The formula is:  

Notation

	=	ith observed response value
	=	ith fitted response
	=	mean response
n	=	number of observations
p	=	number of terms in the model

 Variance inflation factor (VIF)

Used to detect multicollinearity (correlated predictors). VIF measures how much the variance of an estimated regression coefficient increases if your predictors are correlated. Minitab calculates VIF by regressing each predictor on the remaining predictors and noting the R² value. For predictor x1, the VIF is: