Comparable selection is fundamental.
In the traditional “valuation process,” first the “relevant characteristics of the property” are identified. Then in the “data collection” step, “Comparable Property Data” is collected.
The subject is depicted by relevant characteristics (predictor variables), and comparables are chosen to fit that depiction. In the traditional procedure, these are called “elements of comparison.” In the current edition of The Appraisal of Real Estate it states: “This data includes legal, physical, locational, cost, [financial arrangements], and income and expense information . . .” It goes on to relate to supply data, demand data, and estimated future demand.
The above provide good outlines, particularly in historical circumstances where data was sparse and difficult to obtain. Data collection was subjective and subject to availability. Yet it was the most important part of the appraisal process. Appraisers ‘owned’ the data, or had the connections. There was no MLS, no online commercial data, and no electronic public records. A trip to the county courthouse was the norm.
Times have changed.
Picking comparables is still important. But in most areas, complete data sets are available. We have moved from scratching together information to discarding all but the five or six ‘best’ comps. Is this a good data selection model?
What we can do now is work toward an ideal data set. Whether for one of the three traditional approaches to value, for regression, or for the new, Evidence Based Valuation (EBV) – we need to know what gives us the most reliable results. This means, to get the ideal data set, we must figure out two things:
- What is the nature of the right data set? and,
- Exactly how much data is too much, or not enough?
Or as baby bear said: “just right!”
January 27, 2017 @ 2:41 pm
You leave me wanting more. I cannot wait for your next post George. Thank you for sharing and for the great classes you offer.
February 2, 2017 @ 10:28 am
One of the hangovers from reliance on sparse data sets of the early years of valuation is the absence of instruction from the Appraisal societies on objectively measuring comparability. It is widely held that the human mind has capacity to deal with 5 to 7 meaningful items of information at one time, so this cognitive limitation did not hinder picking comps when we only had a handful. Expanding the size of data sets to keep meaningful information, however, requires the use of more objective measures of comparison. There are well developed data analytic tools to perform matching routines and closest comparable selection procedures available to appraisers, but from what I can see little or no instruction in appraisal texts on their use. I, therefore, applaud your efforts to fill the gaping holes in appraisal education with the courses you offer if we are to continue as a meaningful industry in the headwinds of technological change that affront us.
August 30, 2017 @ 7:03 am
Hi Charles,
I am reading some old posts and liked what you said about comparable selection. I try very hard to be objective and do not limit myself in data selection. You said there well developed analytic tools, and comparable selection procedures, are these available and if so where can I review them.
Thanks
Charles Newsom
Certified Appraiser in California
April 21, 2017 @ 4:53 pm
Thank you Charlie. You have been a friend and supporter and inspiration for many years. Charlie is one of the few appraisers who fully applies data science methods and R analytics/reporting software today.
Yes, perhaps your note about “5 to 7” meaningful items, as a cognitive limitation — is a core difference between traditional valuation and the new evidence based methods. Data science is all about how to make sense of 10, 20, or more data points.
October 21, 2020 @ 4:29 pm
We reduce the similarity issue in 5 dimensions.
The several classification algorithms are overviewed in the SGDS2 class.
May 11, 2021 @ 6:37 pm
In a neighborhood where flippers are active, should an appraiser use flipped homes as comps for a home that has not been completely remodeled, flipped and staged for sale?
Do We Need a Neighborhood Section? Part I - George Dell, SRA, MAI, ASA, CRE
February 12, 2023 @ 2:15 pm
[…] In traditional valuation dogma, the next step is picking comps. Comps for overall, comps for income, and comps for land and costs. Pick comps. […]