Page 23 - Regression Guideline for AMC

Screening for incomplete data
• Any proper6es that have values for the characteris6cs that will be included in the regression model that appear to be in error or inaccurate in some way, should be removed before es6ma6ng the model. The distribu6on of each variable should be examined carefully to make sure all values are valid, do not appear to include extreme values, and there are few missing values.
• Key ques:on 2: What was the data screening process used to insure that the data on which the regression model was developed were accurate and complete? Were proper:es with missing or invalid data iden:fied and removed from the analysis and, if so, how many were removed?
• In this case, we iden6fied that the variable acreage was zero for every property except one and we excluded this variable from the analyses. We ran simple tests of frequencies, distribu6ons, and minimum and maximum values on all the variables to insure in-­‐range and valid values. Thus, the property characteris6cs used do not appear to be the problem in this case.
23