Transforming Data For Hotspot Analysis?

Discussion created by jbencina on Jul 20, 2013
Say you had 2,000 locations representing retail locations across the US. Your data includes last year's unit sales for a specific product as a total for each location. Your gross sales range from 0 to 5,000 units however the distribution is not linear. Store 1 sold 5,000, store 2 3,000, store 3 2,100 and so on. Normalizing (min-max) the data results in store 1 at 1.00 and the second store at 0.33.

When working with a hotspot analysis would it be proper to leave the data as-is with a handful of locations over performing the majority? Or would it be appropriate to use a log transformation to make the data more linear before running a hotspot analysis in ArcGIS?