Getis-Ord Gi* Benefits and Strengths

Question

Hello,

Could someone please direct me towards a resource that explains the benefits or advantages of using a Getis-Ord Gi* hot spot analysis over another method, such as a kernel density or cluster analysis.  I need to justify why I used a hot spot analysis, but thus far I am coming up short!

Thanks!

PriscillaCoe · Answer

Hi Cilla,

Thanks for your reply.  I've been digging through some papers, and I largely came to the same conclusions you have.  An excellent piece of citable research that helps explains some of the benefits of the Getis-Ord Gi* (and some of the negative aspects of the Moran's I), is Braithwaite and Li (2007), "Transnational Terrorism Hot Spots: Identification and Impact Evaluation."  The discussion is largely on pages 285 - 287.

I wish ArcGIS were a little more clear on the pros and cons of each tool, especially given how more and more disciplines are adapting geospatial analyses for their research.

Thanks, I'll look at that. It is confusing to decide which tool to use.  I did take a webinar course last week that helped sort it out.

JacobMauslein · Answer

I am in the same boat so let's work on this together.  From what I understand, there are global spatial statistics and local spatial statistics.  The global ones, such as Moran's I or Getis-Ord Gi (without the *), capture global clustering and hotspots for the entire region of your dataset, whereas the local statistics such as Getis-Ord Gi* capture the local clustering and hotspots better.

Kernel density is another way of capturing density or space-use.  Kernel density is non-parametric meaning it doesn't identify statistically significant hotspots.

Hi Cilla,

Thanks for your reply.  I've been digging through some papers, and I largely came to the same conclusions you have.  An excellent piece of citable research that helps explains some of the benefits of the Getis-Ord Gi* (and some of the negative aspects of the Moran's I), is Braithwaite and Li (2007), "Transnational Terrorism Hot Spots: Identification and Impact Evaluation."  The discussion is largely on pages 285 - 287.

I wish ArcGIS were a little more clear on the pros and cons of each tool, especially given how more and more disciplines are adapting geospatial analyses for their research.

PriscillaCoe · Answer

I am in the same boat so let's work on this together.  From what I understand, there are global spatial statistics and local spatial statistics.  The global ones, such as Moran's I or Getis-Ord Gi (without the *), capture global clustering and hotspots for the entire region of your dataset, whereas the local statistics such as Getis-Ord Gi* capture the local clustering and hotspots better.

Kernel density is another way of capturing density or space-use.  Kernel density is non-parametric meaning it doesn't identify statistically significant hotspots.