search

linxihui / NNLM

Nonnegative Linear Models (NNLM) and Nonnegative Matrix Factorization (NMF or NNMF).
Other
76 stars 22 forks source link

Extracting Scores and estimating number of factors #7

Closed gufranca closed 5 years ago

gufranca commented 5 years ago

Hi

I found NNLM an extremely useful tool to find metagene signatures in gene expression data. I was also trying to use the NMF package, which provides extractFeatures() and featureScore() functions to obtain and filter important features for each factor. I was wondering, is there is any similar functions in NNLM for better gene (feature) and metagene selection? I was wondering if there is also some function to help estimating the best number of factors (k)?

I appreciate any help Gustavo

linxihui commented 5 years ago

@gufranca Currently, I have no plan to implement feature scoring. But it should be straight forward after getting the decomposition. For determine k, you can read the paper here for detail, or code example here (the last part). There is not a single function does the work now, but I might do it in the future.