Latent Dirichlet Allocation for dimensionality reduction in clustering

Question

I have a dataset with logs user events. In order to perform clustering on such users, I have derived a set of features of user behavior, including:

• Counts per e