Zachary P Christensen

It may be worth mentioning that any file I've encountered this error with here also doesn't open in anything but Matlab. In other words, I can't open it in Python...

I think this might begin to address a larger problem we have. Although we have the ability to go from a formula to a model, we don't have a way...

After looking at #32 it seems like we want to move towards the formula interface and less so `TableModels`. Would it make sense to have a `getparams` for formulas so...

Related issue is https://github.com/JuliaStats/StatsModels.jl/issues/111, which would be extremely useful.

Thanks for taking a look at this. I'll address the issues if people agree it's a good idea. It was admittedly a messy first attempt to see how practical implementing...

I really like the idea of using `Symbol` for coefnames instead of strings but what would be used for interaction terms? Would we just want to currently developed string name...

Just to add to what @oxinabox said in the OP, terms are also using symbols as names so it seems very odd that the only place we are using strings...

We should probably also address what we'd miss out on if we no longer used strings

> Can you explain in what context does the performance of symbols compared to strings really matters? My point with performance was that using a string to index vs symbol...

Well, if we're not using the traditional table interface then why do we need to use a `String` or `Symbol` for the rowname? Just use a dictionary with the terms.