Mass Appraisal Modeling

Working for the City of Philadelphia, I created a catalog of over 70 regression models that used sales, income and spatial data to estimate values for single family, multi-family, condominiums, large apartment buildings, office buildings, commercial properties, mixed use and multi-purpose properties, and vacant land. I have written both linear and non-linear models, and have also successfully tested gradient boosting models.  

I believe that many counties and municipalities would benefit greatly from incorporating regression modeling into their valuation process. Data quality, cost and expertise have presented barriers to entry for many assessment offices. Using an experienced consultant can help you overcome these barriers and get your office up and running with regression modeling at a cost that can fit any budget. 

If it meets your needs, I can deliver completed models. I have also taught, mentored and trained many modelers, and am happy to work with your staff to develop the necessary skills within your organization so that your efforts can be self-sustained going forward.

Regression models are only one component of the valuation process. You will need data quality analysis to determine if your models can be effectively put into production. In order to apply the models, you also need a projection engine. You will also need tools to review and possibly revise values projected by the models. Getting buy-in from stakeholders is critical! You will need someone who can explain the process to your staff, your county or municipal administration and your constituents. I have been successful in all of these areas. Let me guide you and be your advocate!

Intro to Basic Modeling Fundamentals 101

A presentation that explains how regression models work; what you need to build them and how to evaluate performance.

One Model or Many Models

Examination of the issues and decision points that will allow you to determine the best approach to meet your needs. 

Finding the Secret Sauce

An examination of the use of model specifications, transformations or coefficients from one market to another. Much of the discussion is around the uniqueness of each market, and the need to understand how the relative importance of attributes, market dynamics and the relationship between model constants and adjustments make present difficulties in using model written for one market in a different market.