Skip to content
×
Try PRO Free Today!
BiggerPockets Pro offers you a comprehensive suite of tools and resources
Market and Deal Finder Tools
Deal Analysis Calculators
Property Management Software
Exclusive discounts to Home Depot, RentRedi, and more
$0
7 days free
$828/yr or $69/mo when billed monthly.
$390/yr or $32.5/mo when billed annually.
7 days free. Cancel anytime.
Already a Pro Member? Sign in here

Join Over 3 Million Real Estate Investors

Create a free BiggerPockets account to comment, participate, and connect with over 3 million real estate investors.
Use your real name
By signing up, you indicate that you agree to the BiggerPockets Terms & Conditions.
The community here is like my own little personal real estate army that I can depend upon to help me through ANY problems I come across.
Real Estate Technology
All Forum Categories
Followed Discussions
Followed Categories
Followed People
Followed Locations
Market News & Data
General Info
Real Estate Strategies
Landlording & Rental Properties
Real Estate Professionals
Financial, Tax, & Legal
Real Estate Classifieds
Reviews & Feedback

Updated over 7 years ago on .

User Stats

107
Posts
33
Votes
Brendon K.
  • Real Estate Agent
  • Los Angeles
33
Votes |
107
Posts

How would you draw area maps? For the statisticians

Brendon K.
  • Real Estate Agent
  • Los Angeles
Posted

It occurs to me that we could possibly use data from previous sales to mathematically find average house sales for different areas. Some thoughts:

  • Appraisers use a weighting system to figure out how much the value of a home should be. When you break it down, it's simple comparison using numbers. 
  • Zip codes and neighborhoods do a poor job of isolating different zones of value. You could have three different zones inside a zip code, one for poor houses, one for medium income houses, one for rich houses. Additionally, one type of zone could go into another zone. You could have a rich area going through three zip codes, surrounded by a band of poor areas.
  • The one thing appraisers don't have is a specific mathematical way to determine zone. You look for zones that are similar in how "nice" they look.

What if we were to make some kind of algorithm to draw zones that have nothing to do with zip code or neighborhood? Something to say, "as a baseline, this is how much a 3 bed/2 bath should cost", and it's part of this zone that our program thinks should statically hold a standard deviation of values for other houses? 

We have all the data we need from previous sales, except maybe "does this look like a high end home or not?", which could easily be provided by someone doing data entry at lower wage. How would you go about making such an algorithm, using as little human input as possible?