Cluster analysis?

[sarasustelosantos]

Hello everyone!

If I have a dataset with cities from a country and each city has 10 indicators that result in a final score of maturity, what is the best way to analyse the data considering I want to give recommendations to cities in packages or groups?

I was thinking about doing cluster analysis. Do you think this is the best way to find patterns among groups of cities and get insights on which indicators are of most importance to each group? Or should I use other analysis (within machine learning for example)?

Thank you in advance

[RolandSchubert]

Hi @sarasustelosantos ,

I think, cluster analysis could be the right approach. You have a list of members with attributes and your intention is to create groups, but there is no "general rule" in terms of "if a specific indicator is x, then assign city to group A".  Using cluster analysis is a way to create groups and (in a second step) identify common attributes (e.g. elements in group A have high values for indicator 1, 2, 3, but elements in group B for indicators 4,5,6). The only drawback is, that you have to define the number of clusters in advance, but I would give it a try.

Best,

Roland