Page 22 - Regression Guideline for AMC

Understanding regression mechanics
• The first place to look is always the data that were used to develop the model. Regression models are only as good as the data on which they are based. Data that have been incorrectly entered, have many missing or inaccurately entered values for property characteris6cs, etc. will produce less accurate results. As the old saw goes, “garbage in – garbage out.”
• Correspondingly, the first step in any regression analysis is to examine the underlying MLS data and screen out any proper6es with bad data as indicated by extreme values or even nonsensical informa6on (e.g., -­‐ number of bedrooms at 99, year of construc6on in the early 1800s, etc.).
• It is also important to select proper6es that are not too dissimilar to the subject property in terms of the range of sales prices, year of construc6on, loca6on and other property characteris6cs although we do need some varia6on, otherwise there is nothing to model. The need for some but not too much varia6on in the sample of proper6es to be analyzed (the “Goldilocks principal”) is the balancing trick required for selec6ng proper6es to model.
22