Certified Residential.. can I do mixed use 51% residential How calculated

[George Hatch]

Waterfront in Helendale is usually going to be a PITA. But there are a couple other artificial lake communities, like Canyon Lake in Elsinore or that waterski project out in Indio (I forget the name).

What you need is to talk to a couple of the local brokers who farm that area. They'll know what there is to know.

 

[Tom D]

mixed use will give you double death with the state if you get it wrong. Dear appraiser, what commercial expertise do you have in doing a commercial h&b use, the state asks. After you stop stuttering, the state will find every little mistake you also made to fine and punish you, giving you a painful death. In our state you cannot do mixed use with a residential license.

 

[RCA]

I took an interest in Helendale, CA, because it is a PITA, i.e., hard to appraise area, and situated in the Mojave Desert. The analysis below is a trilogy: MARS (multivariate adaptive regression spline) + GLR (generalized linear regression) + GAM (generalized additive models). For MARS, CRAN earth() is used, for GLR, CRAN glmnet is used, and for GAM, CRAN mgcv is used. These are all 3 very popular statistical regression packages.

For appraisal, MARS is considered number 1 for regression because it provides the appraiser with an interpretable model. GLR is excellent for providing a robust target estimate and is fast, and GAM is good at providing smooth value functions - when it works.

Helendale has been called a PITA to appraise (above), and it is. It has a lot of noise, and many comments could be made about it. It is surprising that earthUI (MARS) and glmnetUI (GLR) performed so well on it; the values they produced were very close. mgcvUI had problems due to noise and other issues in the data. Usually (e.g., Burlingame, CA) mgcvUI also comes up with a target value that is close to the others.

By the way, the value discussed is from a recent sale I randomly chose, and I just gave it an average CQA of 5.00 for the sake of discussion. I have never been to Helendale myself, and of course, what I am presented is strictly just a run of the data through the stat routines. I did categorize water view, which, for Helendale, appears to be an important value contribution. But I have done this without spending any time in Helendale, so I could be off on a few points.

In the link below, there is a comparative analysis followed by the 3 different reports.

Helendale MARS/GLR/GAM Comparative Analysis

I couldn't include the MLS data since it is proprietary, and I didn't want to create altered data. So, no data is included, just the models and graphs and some discussion.

To reiterate, what is stated in the report is not an appraisal.

 

[RCA]

I took an interest in Helendale, CA, because it is a PITA, i.e., hard to appraise area, and situated in the Mojave Desert. The analysis below is a trilogy: MARS (multivariate adaptive regression spline) + GLR (generalized linear regression) + GAM (generalized additive models). For MARS, CRAN earth() is used, for GLR, CRAN glmnet is used, and for GAM, CRAN mgcv is used. These are all 3 very popular statistical regression packages.

For appraisal, MARS is considered number 1 for regression because it provides the appraiser with an interpretable model. GLR is excellent for providing a robust target estimate and is fast, and GAM is good at providing smooth value functions - when it works.

Helendale has been called a PITA to appraise (above), and it is. It has a lot of noise, and many comments could be made about it. It is surprising that earthUI (MARS) and glmnetUI (GLR) performed so well on it; the values they produced were very close. mgcvUI had problems due to noise and other issues in the data. Usually (e.g., Burlingame, CA) mgcvUI also comes up with a target value that is close to the others.

By the way, the value discussed is from a recent sale I randomly chose, and I just gave it an average CQA of 5.00 for the sake of discussion. I have never been to Helendale myself, and of course, what I am presented is strictly just a run of the data through the stat routines. I did categorize water view, which, for Helendale, appears to be an important value contribution. But I have done this without spending any time in Helendale, so I could be off on a few points.

In the link below, there is a comparative analysis followed by the 3 different reports.

Helendale MARS/GLR/GAM Comparative Analysis

I couldn't include the MLS data since it is proprietary, and I didn't want to create altered data. So, no data is included, just the models and graphs and some discussion.

To reiterate, what is stated in the report is not an appraisal.

Note. I did the above without double-checking the data against photos and have never driven to the area. I just found out that the garage space data is wrong, and the model consequently assigns higher values to homes with 0 garages. - Problem: missing garage space values for a number of homes, which were interpreted to be zero garage bays. And, some homes do have 0 garages. ... An example of the importance of knowing the market area. What else is wrong? I can't deal with it, it was just a case of running existing data through MARS. I guess I could check the photos. But I don't have time and I am on "vacation."

 

[Surf Cat]

Note. I did the above without double-checking the data against photos and have never driven to the area. I just found out that the garage space data is wrong, and the model consequently assigns higher values to homes with 0 garages.

This is really interesting. The dude with the best data in the world (MARS, GLR, GAM) gets it wrong because he has never delineated the area or drove the comps. This just goes to show how important not only inspecting the subject is, but driving the comps and viewing them curbside and comparing them to the subject in person.

Kudos to RCA for realizing the error circling back and explaining what needs to take place. An appraiser who has pride in their product.... what a novel concept.

It just goes to show that the new fashion trend of using AI, while being deskbound from another state, no matter how many sources of data you have, cannot replicate the actual viewing of the subject and comparables and having geographical competency in the area.

An example of the importance of knowing the market area.

Hear, hear......