I'm having trouble wrap my head around this one, but not sure.
i) Principal Components Analysis can be used to create a low dimensional projection of the data for use with clustering. Is this True or False?
ii) Cluster analysis can only be performed on continuous variables. True/False?
For ii) I think I can do it on categorical as well as continuous.
For i) I think it is False. Can someone explain why?
For (i) this makes sense...you are effectively projecting data onto a lower-dimensional plane (assuming you don't keep all components).
For (ii) clustering depends on some notion of "distance" (euclidean or otherwise). You can have a dataset that takes values on an integer lattice, yet the clustering approaches can still be used. In addition, you can use measures of correlation (e.g., kendalls tau) even on categorical data and form a similarity matrix from those, so you can "force fit" a lot of things into the clustering framework.