Split out event_study() and plot_event_study()

[grantmcdermott]

Am I right in thinking that the majority of imports in the DESCRIPTION file are just there to cover possible backends in event_study()?

At any rate, I'm unable to install did2s because of a CMAKE dependency issue related to nloptr, which in turn is a recursive dependency of did. It's a long story, but I'm in a constrained cloud compute environment running CentOS 7, which ships some ancient system libraries. Dropping these extra dependencies would facilitate installation of the core did2s package without compromising the main estimation routine. (Please correct me if I'm wrong.)

I think there's a good argument for moving the *event_study functions into their own package, regardless. It would allow you to keep this package focused on its main objective without the dependency headaches. A event_study package would be a pretty natural fit for anyone looking for that set of functions too.

Thoughts?

[kylebutts]

I think I would be fine to this; would you submit a PR to here and https://github.com/kylebutts/event_study?

[grantmcdermott]

Will do. Am i right in thinking that the direct deps of did2s are

• stats,
• methods,
• boot,
• Matrix
• fixest,
• Rcpp,
• data.table, and
• cli

?

[kylebutts]

Yes! I can get rid of cli too; just was being lazy :-)