RE: AI-GEOSTATS: Exploring bivariate/multivariate relationships a s a function of distance
- Thanks for your message, Gaston.
I'm sorry if I was unclear.
Yes - I am definitely going to run a spatial autoregressive model. I am
quite familiar with those approaches.
What I want is to best *explore* how price changes with distance to the
greenway, and particularly find out over *exactly* what range the
relationship is negative.
Basically, there is a change in slope in the scatter plot - over short
distances, the slope of price over distance to greenway is negative, but
then it goes to 0 or is even positive. My interpretation is this is that
the amenity benefit of the greenway is subject to a pretty extreme distance
decay. I want to know more or less exactly what that critical distance is.
The other, maybe less problematic, problem is simply that the decay is not
From: Gastón Pezzuchi [mailto:gpezzuchi@...]
Sent: Tuesday, March 23, 2004 3:05 PM
To: Munroe, Darla K
Subject: Re: AI-GEOSTATS: Exploring bivariate/multivariate relationships as
a function of distance
I am nor particularly well versed in spatial multivariate methods, but I
have found that the best ones are:
a) Geographically Weighted Regression,
b) Spatial Autoregressive Model
c) Spatial Error Model
I do not know which method are you using, or how are you modeling the
space in your problem, and perhaps what I am telling you is something you've
already taken into consideration, but classical multivariate methods aren't
good with spatial data both due to heterogeneity and dependence.
Personally I would consider using GWR methods...
Hope it helps,
----- Original Message -----
From: "Munroe, Darla K" <dkmunroe@...>
Sent: Tuesday, March 23, 2004 1:17 PM
Subject: AI-GEOSTATS: Exploring bivariate/multivariate relationships as a
function of distance
> Hello group,
> I apologize in advance for my lack of geostatistical knowledge. I
> only dabble in these approaches.
> I'm working on a project whose aim is to estimate a real estate
> premium based on proximity to a greenway. The client is a non-profit
> agency who wants to promote "environmental sustainability", and of
> course wants to
> the county that there is a huge positive amenity benefit from being
> closer to the greenway. There is also an incredibly quick distance
> decay; the benefit to the greenway quickly erodes over distance, maybe
> even just a
> hundred meters.
> Problem is, the relationship is so darn complex. First of all, the
> relationship is not monotonic. Presumably, people want to locate
> within walking distance of a greenway, but not necessarily right on
> top of the greenway. Also, there are nice greenways, and bad
> greenways. The
> follows the river, so it runs through some not-nice neighborhoods, and
> in those cases, the greenway is a potential conduit for crime. I have
> looked at distance to parks in this area, and the effect is similar -
> there are some nice parks, and some "drug" parks, so the amenity
> benefit varies.
> Of course, I can control for a lot of these factors in multivariate
> regression, but the problem is, I need to be able to best examine the
> relationship between distance to the greenway and residential sales
> prices to figure out exactly what the distance decay might be, and
> over what
> to best set up this variable in MVR, and to select sample parcels for
> regression. I've done simple scatter plots of price over distance to
> greenway, and it's pretty noisy. The relationship is negative for a
> short distance (i.e., the closer to the greenway, the higher the
> value), and
> basically goes to 0 or slightly positive.
> Can anyone suggest some methods for best identifying the RANGE of
> distance over which this relationship is negative?
> Darla Munroe
> * To post a message to the list, send it to ai-geostats@...
> * As a general service to the users, please remember to post a summary
any useful responses to your questions.
> * To unsubscribe, send an email to majordomo@... with no subject
"unsubscribe ai-geostats" followed by "end" on the next line in the message
body. DO NOT SEND Subscribe/Unsubscribe requests to the list
> * Support to the list is provided at http://www.ai-geostats.org
* To post a message to the list, send it to ai-geostats@...
* As a general service to the users, please remember to post a summary of any useful responses to your questions.
* To unsubscribe, send an email to majordomo@... with no subject and "unsubscribe ai-geostats" followed by "end" on the next line in the message body. DO NOT SEND Subscribe/Unsubscribe requests to the list
* Support to the list is provided at http://www.ai-geostats.org