RCA Method Paper Published

[RCA]

I finally published a paper on the RCA method on Zenodo/CERN. Some may find it interesting.

Residual Constraint Approach (RCA)

This paper describes the new Residual Constraint Approach (RCA), used to achieve higher accuracy appraisals of residential property through the use of R/earth (MARS) and value contribution related constraints.

doi.org

I appreciate reporting errors or omissions.

Notes:
1. This is an official publication for the EU CERN international repository, which will be stored as long as CERN is in existence - probably as permanent as you can get.
2. I will likely overlay this version with a second version by this summer sometime. So, hopefully I will find some errors and make a some additions.
3. - And will write another paper or two.

To Date: No one has reviewed this. But my son-in-law, who has a Ph.D. in Geostatistics from Stanford said he will look it over. Maybe he will even suggest some additions.

Bert Craytor

 

[John Hassler]

Page 4: In the list of counties, change “Santa Rosa”, a city, to “Sonoma”, the county it’s in if, in fact, your experience includes that county.

Edit to add: Not nitpicking, just best to avoid factual errors sooner than later.

 

[George Hatch]

Good for you, Bert. I hope it all works out.

 

[Red Flint]

Thanks Bert, a good read.

 

[Fernando]

Interesting perspective in estimating value based on using physical quantitative variables and then considering residual for the unexplained factor which an experienced appraiser familiar with the location can try to explain for the difference.

 

[RCA]

One problem that bothers me is publishing "real" data. The MLS requires that proprietary IDs be removed, such as MLS IDs. Also, because RCA essentially gives a kind of "grade" in the form of a CQA score to properties, something has to be done to remove or alter other IDs that could ensure a match to some property. So, I actually remove all direct identification such as addresses and parcel numbers as well. Yet, you can't alter the data so much as to give a wrong impression. So, you do what you can do - but if somebody tries to replicate the results with their own download of the data, they will not get quite the same results. Although the patterns will be pretty much the same.

 

[Fernando]

MLS data is proprietary. Isn't it copyrighted and can be taken away anytime by NAR to non real estate agents?
Hate to rely on a model where the data source can be unavailable in the future.

 

[RCA]

Interesting. I uploaded the initial version yesterday afernoon and sent a link to 5 relatives and see 5 downloads.
Then I made a second version with a minor latex correction.
Then this morning uploaded to this forum: Version 2.
Three hours later I decided on another change and uploaded to create version 3 - and updated the above link which was stuck on version 2. But now it is version 2.

What happened: About 32 downloads followed of just version 2, even after I uploaded version 3. So, what happened is someone on the forum sent the version 2 link to one or more others who sent it on, same link. And the count goes up.

I just wonder what happens when I upload to LinkedIn where I have 2500 connections? Or Facebook? Before I do that I better make a new version 4.

By the way, Zenodo doesn't store all those versions, it just stores the changes - and then when someone asks to see a particular version it merges all changes to that version. So, it apparently doesn't take much storage. Except, get this: They store ALL of these versions FOREVER (at least as long as the EU Nuclear Research facility in Geneva Switzerland exists - I assume in some tunnel in the Swiss Alps). All of your stupid mistakes forever.

 

[George Hatch]

I don't think you're selling anything here so the MLS shouldn't have any reason to hassle you for it. Their listing info gets exposed to the public in Zillow and others, including the pics and comments. They're going to get if from the MLS systems before some R&D operator gets it.

 

[Fernando]

I don't think you're selling anything here so the MLS shouldn't have any reason to hassle you for it. Their listing info gets exposed to the public in Zillow and others, including the pics and comments. They're going to get if from the MLS systems before some R&D operator gets it.

MLS and NAR have been forgiven in allowing their proprietary information to be used and given to the public.
If I were MLS, I would charge especially getting some royalties.
Fannie has their data from unknowing appraisers.
RCA method is using data from MLS which may be a problem later on if used for profit.