String Index for DataFrames

Specific Domains Data
1

I am using the DataFrames package, and I would like to set a string column of it as the index. For example, let x =

│ Row │ name   │ val│
│     │ String │ String │
├─────┼────────┼────────┤
│ 1   │ A      │ 1│
│ 2   │ B      │ 2│
│ 3   │ C      │ 4│
│ 4   │ D      │ 3│
│ 5   │ E      │ 4│
│ 6   │ X      │ 5│

Then, I want something like x[“A”, :val] to return 1. Currently, I have to do the following

@where(x, :name .== "A")[1, :val]

which is bulky and unclear that the @where should only return one row. Is there anything like pandas set_index or an alternative? I’m having trouble understanding setIndex! in the DataFrames documentation, but it states that indices must be integers in any case.

2

Unfortunately currently DataFrames.jl does not allow indexing. setindex! is a function defined in Base and is used for setting values in a DataFrame. What you can do is:

filter(row->row.name == "A", df).val

or

df.val[df.name .== "A"]

or

df[df.name .== "A", :val]

or some variant of these.

There is an open PR https://github.com/JuliaData/DataFrames.jl/pull/1908 that after merging will allow you to do the following:

gd = groupby(df, :name) # this kind-of creates an index on df by grouping it
df[("A",)].val # get group defined by "A" as a SubDataFrame and fetch :val column from it
4 Likes