Support for SubDataFrame and other DataFrame Types

[rcnlee]

The DataFrames package may produce derivative types depending on the operations performed. For example, if you take a view into a DataFrame, you get a SubDataFrame. A groupby gives a GroupedDataFrame, etc.

Plots.jl can be easily modified to support all of these by changing the function signatures to accept the AbstractDataFrame type. In particular, the functions in plot.jl dataframes() can be modified like this:

@eval function createKWargsList(plt::PlottingObject, df::DataFrames.AbstractDataFrame, args...; kw...)
    createKWargsList(plt, args...; kw..., dataframe = df)
  end

  # expecting the column name of a dataframe that was passed in... anything else should error
  @eval function extractGroupArgs(s::Symbol, df::DataFrames.AbstractDataFrame, args...)
    if haskey(df, s)
      return extractGroupArgs(df[s])
    else
      error("Got a symbol, and expected that to be a key in d[:dataframe]. s=$s d=$d")
    end
  end

[ErwanPiR]

Hi,
I wonder if this question has been adressed, because when I try to plot SubDataFrame or GroupedDataFrame, I get the error :
ERROR: No user recipe defined for SubDataFrame{Array{Int64,1}}

Thanks.

[mkborregaard]

You can't. See StatPlots for plotting of DataFrames

[ErwanPiR]

Ok, thank you. I already check StatPlots but it's the same error :)

[mkborregaard]

Yes because there is no recipe for DataFrames. If you have a use case that isn't covered by the functionality in StatPlots you're welcome to open an issue with the example.

[ErwanPiR]

Thank you. I open an issue in StatPlots to explain the case.