Set of operations to benchmark

[ChrisRackauckas]

I think we should come up with a pretty comprehensive set of operations to benchmark. Is there a standard list for the field? Probably something with some groups, joins, etc. @davidanthoff or @explodingman might have some good resources?

[xiaodaigh]

Yeah, can test

leftjoin, innerjoin, outer join, and antijoin.

transpose, melt etc

Basically we can look into dplyr and data.table operations and just add one operation at a time.

[ChrisRackauckas]

Here's something to look at and track: JuliaData/DataTables.jl#17

[ChrisRackauckas]

Could we get them generating one big table like https://github.com/Rdatatable/data.table/wiki/Benchmarks-:-Grouping#code-to-reproduce-the-timings-above-? We can do Pandas through Pandas.jl and data.frame + dplyr through

[xiaodaigh]

I've tried before Rdatatable/data.table#974

I am actually interested in benchmarks using real-life data instead of purely relying on synthetics

[ChrisRackauckas]

I am actually interested in benchmarks using real-life data instead of purely relying on synthetics

A mix is usually a good idea.