Respected all,
I am working on the classification of pollen on the basis of locality. Using PCA, I have got a good differentiation and classification of samples. Now, should I also go for using the cluster analysis and LDA for supporting the results of PCA or not? If yes, then what is the basis of further use of cluster analysis and LDA?
Thank you all.
The goals of PCA, LDA and cluster analysis are different. PCA breaks up the variables into groups on the basis of their contribution to the overall variance. The principal components are interpreted in terms of the original variables. Often a few components are enough to explain more than 90% of the total variability. The LDA is to put the observations into categories identified a priori and cluster analysis is to identify closely related variables. The categorizations are not necessarily related for the three techniques. The techniques are independent. There is no logical sequence from one to another. There is no basis for arranging them in any particular sequence.
Mamta
What are the basic hypotheses you are trying to test with your data? Dr. Etuk's explanation concerning the application of these different analytical approaches is spot on. You would use cluster analysis for group identification which is often coupled with other visualization approaches like multidimensional scaling. If you already have localities then perhaps a discriminant approach coupled with MANOVA is the best approach. Can you describe your design a bit more?
I'm sure you will find the answer here
Article Linear discriminant analysis: A detailed tutorial
Article Principal component analysis - a tutorial
Hello every one, I would also know the difference in efficiency between LDA, PCA CCA, RDA, NMDS and cluster analysis for geographical origin discrimination of agro-products?
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India