I am using ArcPro to analyze disease trends in my state. I am looking retrospectively at 6 years of data. I want to identify hotspots at the census tract and county level. I am performing this analysis on several disease serotypes to compare geographic distribution. I am a bit overwhelmed by the choice of parameters. Inverse distance, fixed distance, nearest neighbors. I am hoping to turn this project into a paper. Most papers I've read use zone of indifference or inverse distance band method. The disease I'm looking at does not have a known distance of transmission. It can be random.
I've done some comparisons using the optimized hotspot analysis and compared it to my results from the hotspot analysis with various parameters, and surprisingly, the optimized analysis matched the results for the fixed distance band, which generated larger areas of hotspots, while the inverse distance band (at the distance recommended by the Global Moran's spatial autocorrelation) yielded more sporadic, isolated hotspots.
For the purpose of the analysis, I'm having a difficult time deciding if 1) hotspot analysis is better than optimized hotspot? 2) which parameter settings to use? I've read the documentation, and it sounds like inverse distance, zone of indifference, or nearest neighbors would be better than fixed distance band, however, the fixed distance band, zone of indifference, and even the nearest neighbors seem to match the results from the optimized hotspot analysis, (not the inverse distance) which was not what I was expecting. The fixed distance and zone of indifference in particular are closest to the results from the optimized hotspot analysis.
I'd appreciate any advice or further explanation. This is new territory for me on my GIS journey.
