rsnirwan
rsnirwan
By 'any collection of both' is meant that it runs either on a collection of `pandas.DataFrame` or a collection of `pandas.Series`. Maybe a better wording would be: > `PandasDataset` is...
No, it isn't much work. But I would propose not to do it anyway.. Imposing same structure on all elements of the collection seems to be nicer. If someone comes...
I wrote down my thoughts on this in another issues I'll link below.
> Can we have some motivating examples where we can try out the different approaches? For me, the proposals are a bit hypothetical. The question seems a bit too general....
Even though, I am using them everyday at work, it's not easy to find publicly available time series in wide format (see an artificial example below). So, let me answer...
Hi stouffer, yes, you are right. We overparameterize the matrix by `q` parameters. The reason for that is the sampling of each `v` from the unit sphere. It's not easy...
1. You might get non-convergent chains in the case of our proposed parameterization. This is due to the fact that there is no way of continuously mapping a sphere S^(N-1)...