These methods are very powerful, the JMP help files can provide more information about the clustering criteria and how they actually work. But mainly, I'm curious about how? (in a practical sense) the hierarchical clustering method (Ward) is identifying the separation between the countries. How do we draw this back down the the actual data itself in a practical way? I guess another way of asking this would be: What does it mean to "identify the underlying structure" in your data? What does it mean to have "the same birth and death rate characteristics?"

@JMPStatisticalDiscovery

3 жыл бұрын

The underlying structure in data for this particular example refers to the fact that given the limited knowledge that is contained in this dataset which countries have similar characteristics. For example, in this dataset instead of birth-death rates If we had only surface area information then the US and Canada will be in a similar cluster (en.wikipedia.org/wiki/List_of_countries_and_dependencies_by_area) ; however if we had population density instead then Canada will be in similar cluster with Iceland ( en.wikipedia.org/wiki/List_of_countries_and_dependencies_by_population_density )

@PatrickGiuliano

3 жыл бұрын

​@@JMPStatisticalDiscovery Great thanks for that clarification. That makes good (practical) sense to me. It's sort of like correlation but of course the two concepts and statistical methods should NOT be conflated. The clustering approach, in particular for Hierarchical in this example shown, is principally not working by a means of calculation of Pearson correlation (or covariance), but rather, uses different math to determine cluster aggregation (e.g. Ward, Average, Centroid). If I'm not correct on this please do let know.

Really weird that JMP doesn't comprehend that users are looking for a quick intro on the potentials/limitations of JMP, not a poor tutorial about clustering methods, etc.

@sportsguru3786

5 жыл бұрын

Right this is way too in depth. Makes me more confused