Addtional Road Buffers
Robert Denne has made the following suggestion:
Rob, You may already have worked this out, but if not I thought you might like to know that a gradient can be a worthwhile addition to the major-road effect I reported previously. My initial discovery was that a 100 foot buffer was associated with a notable diminution of parcel values, which Takuya Ando later confirmed and incorporated in your modeling software using a 300 ft buffer. I have since explored both a 100 foot buffer, which yielded a -.094 coefficient on a virtually identical log-log model, and a 101 to 300 foot buffer in the same model, which generated a -.036 coefficient. The top lines of the new SPSS coefficient table follow:
Model Unstandardized Coefficients Standardized Coefficients t Sig.
B Std. Error Beta
1 (Constant) 7.182 .076 94.849 .000
LnHDSF .086 .004 .105 19.386 .000
LnBldgSF .628 .009 .498 67.919 .000
MajrRd100FtBuff -.094 .005 -.055 -17.472 .000
BufFlg101300 -.036 .004 -.030 -9.542 .000
For comparison, the old 100 foot single buffer generated a -.124 coefficient with a t value of -27.946. I’ll try to master your R modeling approach, but in the meantime all I can conveniently provide are SPSS-oriented results.
I plan to turn next to exploring whether data are available to do similar analyses with railroad stations and lines – benefits as well as disbenefits, and also green space, particularly park land and suburban high-tension-line corridors, although I’m far from sure the data for either are readily available.
-
Verify that including additional buffers improves modeling outcomes. -
Create new models based on highest performing models to use these additional buffers.
Regards, Bob