Hot Spot Clusters

[LaurenSmith]

Hi,

This is probably very simple to answer. I am running the Hot Spot Analysis for different years to examine any clusters of diseases. I have followed the on line video instruction and was wondering once I have distance band for one year do I use the same for all years or should I run the autocorrelation for each year to find the best distance band? I assume I do that but wasn't sure if it would mean the clusters are not comparable or something similar.

Thanks

[OliviaPowell]

Hi,

I have just found the answer myself to this.
The Distance band depends on your cell size which depends on the size of the area you are working on.
e.g. you're working on a specific administrative boundary - county for instance.

1) You need to create a fishnet / or grid.
From Spencer chainey's methodology (Crime mapping), it appears that your grid size needs to be 100 times smaller than your smallest size of the area you're working on.
E.g. County is 20000 meters by 40000 meters. Then your cell size of your grid should be 20000/100 = 200 m by 200m.
in ArcMap 10, you need to use Toolbox > Data management Tools > Feature class > Create Fishnet

2) Then use your Hot Spot analysis (Getis-Ord) tool
Your distance is the diagonal of your cell. To calculate this, you need to use the Pythagoras formula

It's all explained in a presentation I have seen in a conference.
follow this link => Class 3 D
http://www.ucl.ac.uk/jdi/events/int-CIA-conf/Abstracts/ICIAC11_Classes3
Videos are also on you tube.

Good luck

O