SSE,SSR,MSE,MSR

[Alex_sad]

Hi
How can i reach to sum of squared and mean square of regression and residuals in a regression model?
In another term, i want a table like what can be created in SPSS for ANOVA...
I want to use SSR and MSE in a equation and i need these figures.
Help me plz

[EViews Gareth]

The regression output shows SSR at the bottom.

If you hit the forecast button, and forecast over the estimation sample, you'll see MSE reported. Or you can calculate it yourself from SSR.

[Alex_sad]

yes there is just SSE(for residuals).
How can i calculate SSR(for regression) and MSE?

[EViews Gareth]

What definition of SSR are you using?

[Alex_sad]

total sum of squares = explained sum of squares + residual sum of squares
in this equation i want to find two variables in the right hand sight in eviews results.

[EViews Gareth]

SSR under that definition is shown on the equation output. It is labeled as "Sum squared residuals".

[Alex_sad]

yes it is, but what about the other?

[EViews Gareth]

Sum of squared deviations? Just open up your dependent variable and click on View->Descriptive Statistics->Stats table.

[Alex_sad]

i attach an image!
in anova table there is three row: Regression, Residual and Total, i want the number that is in front of regression (in this example 274405.373)
Thank you so mush

[EViews Gareth]

Unfortunately that picture doesn't describe what that number is. What is the calculation you want?

[Alex_sad]

SST = SSR + SSE
SST= ∑〖( y - y ̅ )〗^2
SSR= ∑〖( y ̂ - y ̅ )〗^2
SSE= ∑▒〖( y - y ̂ )〗^2

[EViews Gareth]

Under those definitions, SSE is shown on equation output (as sum of squared residuals). SST is shown by opening up the dependent variable series and clicking on View->Descriptive Statistics->Stats table.

SSR you'll have to calculate.

[adrangi]

I think you're asking for SSR (sum of squared of explained variation) and SST sum of squared of the total variation. Normally Excel will simply show the three in the ANOVA table where SST=SSErrors+SSReg, from there you can verify R-sq=SSReg/SST. Unfortunately I have not been able to observe SST and SSReg (as defined above) directly in an Eviews output. Here is my way:

On the command line (above your computation area) type the following:

scalar stdevy=sddep this gives you the standard deviation of the ys.
scalar vary=stdvy^2 this gives you the variance of the ys from the sample
scalar ssty= vary*(n-1) where n is the number of observations. This gives you SST of the sample as I defined.
scalar sserrors=@ssr
scalar ssreg=ssty-sserrors
scalar rsq= ssreg/ssty

Hope this helps. May look like a lot of typing but goes quick if you don't have typos. Best, B