search

stephenslab / fastTopics

Fast algorithms for fitting topic models and non-negative matrix factorizations to count data.
https://stephenslab.github.io/fastTopics
Other
77 stars 7 forks source link

Implement function to estimate loadings in held-out samples #4

Open pcarbo opened 4 years ago

pcarbo commented 4 years ago

Fix F, estimate L for the held-out samples, then evaluate the log likelihood of the held-out samples.

In sklearn, the API is fit(X) to estimate L and F, and transform(X) to estimate L for some new X (fixing F). I'd probably not call it predict, since I would expect such a function to return the estimate of Lambda = LF'.