Spatial Statistics of Categorical Data

[KendraRodgers1]

Hello out there, hopefully some of you have recovered from the UC and are still checking out the forums.

I am trying to wrap my head around the spatial statistics toolbox, and was attempting to do some exploratory regression work today on an archaeological data set....when I realized the data I want to look into further is all categorical or binary.  From what I can tell regression is not the way to go about this for this type of data.  Is this correct?  Am I essentially now restricted to cluster or point patterns?  I don't believe I can even use Global Moran's I ... however I'm have about 9 weeks of beating my head against my desk when it comes to spatial stats.  So that could be wrong.  Any pushes towards reading material or toolboxes geared more towards categorical and/or binary would be much appreciated.

Thanks,

Kendra

[DanPatterson_Retired]

You really don't indicate what you intend to use spatial statistics for.  Perhaps conventional nonparametric statistics is all you really need (ie chi-squared testing etc).  A lot of the tests rely on ratio or interval data, and I am not sure that traditional Joins Count Statistics can be emulated within one the other spatial tool options.  If you have some questions that you would like to address, it would help individual formulate better suggestions.

[KendraRodgers1]

Hi Dan,

I was hoping to look at the locations of one kind or archaeological site or type of event compared to the other events in the area for trends.  For example, event B occurs near A, but event type C and D do not occur by A.  Right now I have identified all they categories (A-G,etc) and was hoping to use exploratory regression tool to understand if B-G have any influence on A.  I think some sort of spatial stats is still in order as I'm trying to assess if events occurred in specific places and in spatial relation to other specific events in the landscape.  It really is a form of settlement analysis. Hope that gives you a better picture of what I'm trying to accomplish.  Thank you.

[DanPatterson_Retired]

Not limited to... tests for nominal/categorical data

If you wish to pursue regression, then use Google to search using... "regression analysis" "categorical data" as key phrase searches.  There are scholarly articles if you have access to online journal acquisition.

Spatial statistical analysis for categorical (ie nominal data) within GIS doesn't provide many tools although implementation of "joins count" statistics can be found.  Scholarly papers will provide some reference to others.  Keep us posted with findings.

[KendraRodgers1]

While I'm not limited to categorical data, I want to include some categorical data for my research.  I am currently looking into GWR 4.0 as a possibility to use for my analysis.