• Welcome to AppraisersForum.com, the premier online  community for the discussion of real estate appraisal. Register a free account to be able to post and unlock additional forums and features.

Regression Coefficients

Status
Not open for further replies.

Valueseeker

Junior Member
Joined
May 19, 2016
Professional Status
Certified Residential Appraiser
State
Massachusetts
I don't claim to be a pro in multi variable regression. I'm learning. I want to learn more. This is a question for those who are well versed in it.

I ran a regression with approx 100 sales. Using a comparable as a test to check how close this predictor is, I found that the model was almost right on value. I've never used any of the values in my adjustments. I'm referring to the coefficients and that is my question. Are the coefficients in a regression model the value in which the statistics support? I've never adjusted 32k for a garage bay. Obviously it's off. But the pval seems to be useable no?

Anyhow. I don't want to chalk this up to why regression can mislead your adjustments but a better understanding as to how to properly and appropriately consider the data it spits out.

So. Coefficients... the suggested adjustment amount? If so, What else is required to make them realistic amounts?
 

Attachments

  • IMG_8792.PNG
    IMG_8792.PNG
    347.1 KB · Views: 40

Randolph Kinney

Elite Member
Joined
Apr 7, 2005
Professional Status
Retired Appraiser
State
North Carolina
Look at your p-values (good if less than 5%, higher means you can reject it). Also know which variables are interacting with GLA like bedrooms, bathrooms, and garages. They will have unrealistic coefficients for adjustments.
 

Terrel L. Shields

Elite Member
Gold Supporting Member
Joined
May 2, 2002
Professional Status
Certified General Appraiser
State
Arkansas
I would never use land as a variable. Land does not depreciate. Regression will capture functional depreciation.
 

Terrel L. Shields

Elite Member
Gold Supporting Member
Joined
May 2, 2002
Professional Status
Certified General Appraiser
State
Arkansas
boot acres out apply sales for land adjustments. Put variables in /out and see how that changes.
 

Elliott

Elite Member
Joined
Apr 23, 2002
Professional Status
Certified General Appraiser
State
Oregon
VS said, Using a comparable as a test to check how close this predictor is, I found that the model was almost right on value.

So you tested one comparable? I'm willing to guess if you test all 100 comparables, 67% will be within one standard deviation. Welcome to the wonderful world of statistics. Zestimates have a similar record of being 'right' and drive Realtors crazy and confuse HOs on a monthly basis.

Now go forth into the real world and on your next appraisal, apply your factors, don't fiddle with it, and send in the appraisal. Let us know what happens. Yeah, I don't adjust $32K for a garage either.

PS--Heard a good analysis of why Hillary lost. It was because the polls were so wrong. She would have been able to touch more donors for another billion dollars if it was close, but because she was a shoe in they didn't give her more money.
 

Valueseeker

Junior Member
Joined
May 19, 2016
Professional Status
Certified Residential Appraiser
State
Massachusetts
I understand. At the end of the day I would even like my coefficients to be close to what I normally adjust for. Why? I guess it would satisfy myself into thinking I've further supported my adjustments from matched pairs. Is it ever possible to tweak it in a way so the coefficients are basically what we(me) adjust for?
 

bart nathan

Member
Joined
Jun 9, 2005
Professional Status
Certified Residential Appraiser
State
California
I would try to run again a couple of times and remove some different variables Bed, Baths and fireplaces and see what happens. You also need to filter some of you sales to and toss out some of the outliers really high and really low sales. After that you'll probably have something more reliable. Multiple regression will work better using less variables. Example would be if your only looking for a GLA and Lot size adjustment try only using sales with only those differences or just use linear regression if your only looking for one variable.

Multicollinearity definition below is probably why your garage and/or some of your other predictors are not reliable and also what the software providers selling these programs don't want you to know.

"In statistics, multicollinearity (also collinearity) is a phenomenon in which two or more predictor variables in a multiple regression model are highly correlated, meaning that one can be linearly predicted from the others with a substantial degree of accuracy. In this situation the coefficient estimates of the multiple regression may change erratically in response to small changes in the model or the data. Multicollinearity does not reduce the predictive power or reliability of the model as a whole, at least within the sample data set; it only affects calculations regarding individual predictors. That is, a multiple regression model with correlated predictors can indicate how well the entire bundle of predictors predicts the outcome variable, but it may not give valid results about any individual predictor, or about which predictors are redundant with respect to others."
 

Elliott

Elite Member
Joined
Apr 23, 2002
Professional Status
Certified General Appraiser
State
Oregon
Newton got hit on the head with an apple and had to develop a system of mathematics to describe it, along with the orbits of planets. Einstein needed to use mathematics to describe the forces of the universe. The origins of statistics are less noble, Pascal developed statistics to assist gamblers who might be raided and need to know how to divide up a pot. Its predictive powers are over rated.
 

Valueseeker

Junior Member
Joined
May 19, 2016
Professional Status
Certified Residential Appraiser
State
Massachusetts
I would try to run again a couple of times and remove some different variables Bed, Baths and fireplaces and see what happens. You also need to filter some of you sales to and toss out some of the outliers really high and really low sales. After that you'll probably have something more reliable. Multiple regression will work better using less variables. Example would be if your only looking for a GLA and Lot size adjustment try only using sales with only those differences or just use linear regression if your only looking for one variable.

Multicollinearity definition below is probably why your garage and/or some of your other predictors are not reliable and also what the software providers selling these programs don't want you to know.

"In statistics, multicollinearity (also collinearity) is a phenomenon in which two or more predictor variables in a multiple regression model are highly correlated, meaning that one can be linearly predicted from the others with a substantial degree of accuracy. In this situation the coefficient estimates of the multiple regression may change erratically in response to small changes in the model or the data. Multicollinearity does not reduce the predictive power or reliability of the model as a whole, at least within the sample data set; it only affects calculations regarding individual predictors. That is, a multiple regression model with correlated predictors can indicate how well the entire bundle of predictors predicts the outcome variable, but it may not give valid results about any individual predictor, or about which predictors are redundant with respect to others."
Right. If I add sea salt and table salt to my meal, my tongue can only tell that the food is salty. Same with the model.
 
Status
Not open for further replies.
Find a Real Estate Appraiser - Enter Zip Code

Copyright © 2000-, AppraisersForum.com, All Rights Reserved
AppraisersForum.com is proudly hosted by the folks at
AppraiserSites.com
Top

AdBlock Detected

We get it, advertisements are annoying!

Sure, ad-blocking software does a great job at blocking ads, but it also blocks useful features of our website. For the best site experience please disable your AdBlocker.

I've Disabled AdBlock
No Thanks