Regression Analysis for GLA

[cushie0105]

Is it possible to extract an adjustment from the market for the GLA $ per sqft adjustment? I notice I can’t extract GLA or garages for some reason.

Is there a trick to extracting the GLA adjustment?

 

[TerryRohrer]

Graph Sale Price (vertical axis) and GLA (horizontal axis) in Excel, then add best fit line.

How are you trying garage stalls?

 

[cushie0105]

For garages I do township wide jonlakefronts 1500-3000 sf. Then get the average for no garage. Then get the average for 1 car then 2 car.

Everytime I do it I get a $25,000 difference

 

[Jerry Bone Jr]

Graph Sale Price (vertical axis) and GLA (horizontal axis) in Excel, then add best fit line.

Yes, but instead of sale price, use the adjusted sale price on the y-axis.
After all other adjustments are made, use the ASP.

 

[alebrewer]

The actual 'adjustment' (if you're doing a bivariate regression) would be the number attached to the 'x' in the best fit equation. Remember that the equation for a line is Y = mX + B, where B is the Y intercept when x = $0 (for obvious reasons). Let's say your line equation is: y = $100,000 + $102.3x. The actual adjustment factor, then, would be $102.3.

 

[RCA]

Is it possible to extract an adjustment from the market for the GLA $ per sqft adjustment? I notice I can’t extract GLA or garages for some reason.

Is there a trick to extracting the GLA adjustment?

With MARS regression, you can extract adjustments for many features - as long as they are providing a significant contribution to the sale price.

NOTE:

1. Regression does not directly give you adjustments - it gives you value contributions. Generally speaking, you should take the value contribution for the subject minus the value contribution for a given comparable to get the adjustment. This method also handles non-linear regression models such as MARS, that is models where the price/sf will change between smaller and larger GLA homes.

2. You can't get a good adjustment on one feature alone, you need all of them because there is so much collinearity between different features. For example, GLA is invariably co-linear with Room Counts - larger homes also have more rooms. GLA is often co-linear with lot size and stories. And "garage bays" is usually co-linear with GLA. Another example is that large homes are usually worth more if they have an appropriate number of bathrooms.

... There is so much more to it.