Spatial Interpolation on Duplicate Points

685
4
02-09-2011 01:04 PM
JohnBarr1
New Contributor
Attempting to create a surface representing property values. 
Single family homes (which are) - 1 value : 1 parcel - is simple enough, but would like to include condos, which are Many values : 1 parcel.

My data sets are:
-Polygons for Parcels (Parcel ID)
-Table with attributes: Account ID, Parcel ID, Account Value.

My methodology is to convert the Parcel Polygons to points, then join to table on Parcel ID to create a single point with Value for a single family home; however, the multiple condos values are forced to relate to a single point.

Because of condo values being coincident to the single point location, my understanding is that this is problematic.  The solution I have in mind is to place a number of random points equal to the number of units within the Condo within the region defined by the parcel polygon.  That way I have represented all values by a point within the parcel extent that it belongs to.

I am presuming that I will be using IDW to derive the surface, and therefore would like to ask if there is a tool to force the points to be distributed to an region that is defined by each parcel polygon.

Or is this a good example why another method to derive the surface should be used, ie Kriging?

Thanks.
0 Kudos
4 Replies
JohnBarr1
New Contributor
Bill; thank-you for your response.

I agree that the application of the technology for deriving property value is obscenely crude; but the end goal is simply to provide a qualified visualization of assessed (land and improvements) value on a large scale.  With the aim solely being to provide technology eye-candy, and not a means to extrapolate value from the model; wouldn't the generated surface accomplish this?

While I appreciate the thought and expertise in your response, the intent of my post was to discover how best to approach interpolation when the data set includes coincident points with varying values; I understand that IDW will select only one of the values, and which one it picks is unpredictable �?? and Splines will fail when more than one value is given for a common location.   I am therefore left with the choice of scattering the points within a region, or using Kriging which will conduct a statistical summary of the points before processing the surface. 

I�??d like to confirm that my understanding of Spatial Interpolation is correct, and see if my scattering solution is feasible.
0 Kudos
SteveLynch
Esri Regular Contributor
0 Kudos
JohnBarr1
New Contributor
As this has been my first post to the forum, and my first few weeks using Spatial Analyst I have the following observations:

1) It is very easy to produce a surface.
2) A 3d image of a surface is an impressive visual tool.
3) Non-technical executives value the impact of the image more than the validity of the result.

4) Any surface, because it is an interpolation, includes some degree of error.
5) There integrity of the result is limited by primarily, the expertise of the operator to select the most suitable parameters for the tool, and secondly the ability of Spatial Analyst to provide an appropriate tool for the topic (Elevation, Temperature, Value, etc).
6) It must be disconcerting for experts in this field to have such a powerful tool be made available to neophytes who wield this new tool with ignorance of the integrity of the result, especially when their work is readily consumed without question by those looking for the sizzle and not the steak.

7) Moving from Vector to Raster, is like putting an algebra instructor into a philosophy class.  Where once they have been accustomed to dealing with cold hard facts, they are now faced with making judgements based on a comparatively subjective basis, and every decision they make introduces more and more chance of error into the solution.  We have moved from discrete "right and wrong" to "better and worse".  However, this uncertainty can be managed by applying a statistical methodology.

😎 Questions lead to answers, which lead to more questions.

Thanks.
0 Kudos
DanPatterson_Retired
MVP Emeritus
John
Good summary...distribute amongst your fellow students and keep a copy for yourself should you choose to go into teaching 🙂
0 Kudos