Emerging hot spot analysis and non-continuous time steps

[Anonymous User]

I have 10+ years worth of data that I would like to run an emerging hotspot analysis on. My time steps are not continuous (e.g. covers all years from 1998 to 2010 except for 2006). I know I can physically run the analysis, but I was wondering if there were any assumptions that I would be breaking that means I really shouldn't run the analysis (beyond acknowledging that there is a gap in the time intervals)? 

Many thanks

[LynneBuie · Accepted solution]

Hi Sam Andrews‌,

The table in this documentation page shows you what each of the Emerging Hot Spot Analysis patterns actually mean. You can imagine that many of these might change if a time step is missing - for example, a New Hot Spot is "A location that ... has never been a statistically significant hot spot before.". Perhaps this location was a hot spot in your missing year, hence this result might be wrong.

There isn't really a perfect solution but I see two options - one, as you suggested, simply acknowledge that there is a gap that could impact the results. Two, is to impute the values of the missing year with sensible values based on the values of neighboring years. I would do this by creating your 2006 features with null values, then filling these values using the Fill Missing Values tool with the Fill Method as Temporal Trend. This way you get an imputed value at each location for 2006 that adheres to the general trend of the values in each location.

Regards,

Lynne

[Anonymous User]

Thank you for this - it is very helpful. I feel much more confident in selecting the best option.