Juniper Simonis

EDIT: this unfortunately strips the metadata / breaks internal links, so strike that idea! this is something i've hit on recently, as well, and after trying to implement something and...

> > > Won't this step overwrite the PDF metadata? unfortunately, yes :( back to the drawing board

this should be accomplishable with the multi-LDA_TS functionality concept #149 and specifically #146 we can build simple "identity" and "identity but with error" models that are functionally 1-topic LDAs

this should be accomplishable with #149 and incorporation of replicated runs of a given `LDA_TS` (perhaps replicates to show variability or variations of a cross validation schema)

also tidy the matrix situation, see https://github.com/weecology/LDATS/pull/142#issuecomment-568784202

we'll want to have some element at the top level and perhaps at lower levels in the model objects that indicates which data are used/not out of a whole data...

totally good idea on that approach. it'll definitely require more computation, tho (like most things with this method), so is another benefit of having more efficient coding for the computation...

ref: #119

also, conversion back to user-input times will be nice, ref #123

a specifically important instance of this general issue is the non-integer times, as discussed in #126