I have a regional count data of disease incidence. First, I calculated empirical Bayes estimate of standardized morbidity rate. Then I performed global Moran's I to check for spatial autocorrelation of standardized morbidity rate. The result showed that spatial autocorrelation is not significant.

Should I still perform detection of the location of a spatial cluster?

  • Well, if there is no spatial structure in the data then any clustering approach that relies on said structure will be invalid. If you have variables related to the data then you can use something like k-means or Agglomerative Hierarchical Clustering, and you can even include [X,Y] coordinates as a sort of spatial constraint. There is even a function hybrid.kmeans in the R package spatialEco implementing a hybrid approach that uses Hierarchical Clustering to find the cluster centers for a k-means. – Jeffrey Evans Aug 27 '20 at 15:59
  • This is hard to answer without knowing anything besides that there is count data for diseases. Are these data held in polygons? If so, how many polygons are we talking about? Also significance of Moran´s I depends on the number of samples. I guess you have enough, but why not try a correlogram? I would never trust Global statistics and always check 1,2,3 neighbors and so on. – Jens Sep 2 '20 at 20:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.