If I want to test for statistical assumptions for Euclidean distance, Manhattan, HCA and LDA , is there any software to use ?how do I test each of them? My objective is to determine which statistical method is suitable for drug comparison?
What does Euclidian distance have to to with drugs? And Manhattan? And what do you mean by HCA and LDA? Not everyone comes from your exact biotope where these words are used in a specific context.
I presume that, in your query, HCA means hierarchical cluster analysis. But, to be clear, what does LDA mean?
1. Linear discriminant analysis?
2. Latent dirchlet allocation?
3. Something else?
WIthin HCA, the choice of distance metric can and often will make a difference in the resultant cluster solution options. The same is true for decision rules about stopping the clustering, whether to standardize variables, agglomerative vs. divisive methods, and a host of other decision points. I'm not quite sure which assumptions (orthogonality? interval strength scale? something else?) you're eager to evaluate, so perhaps you could elaborate your post somewhat. Perhaps, as well, share the specific research question(s) you're trying to address.
If you did, I suspect you would be much more likely to receive more focused recommendations.
In addition to Prof Morse's post you may find the attachment useful to you. it's available in the z-library. All of the software is available in R, package cluster Best wishes David Booth