Permutedims not working on DataFrame

General Usage
1

Hi all,
I am trying to transpose a DataFrame, following what is explained here: Functions · DataFrames.jl

When applying the ‘permutedims’ function, I get an error that says that this function cannot take a DataFrame as an input.

julia> typeof(df_all)
DataFrame

julia> df_transpose = permutedims(df_all,1)
ERROR: MethodError: no method matching permutedims(::DataFrame, ::Int64)
Closest candidates are:
  permutedims(::StridedArray{T, N} where {T, N}, ::Any) at multidimensional.jl:1494
  permutedims(::FillArrays.AbstractFill, ::Any) at /home/maf190004/.julia/packages/FillArrays/7oBjc/src/fillalgebra.jl:21
  permutedims(::SparseArrays.AbstractSparseMatrixCSC, ::Any) at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.6/SparseArrays/src/sparsematrix.jl:1068
  ...
Stacktrace:
 [1] top-level scope
   @ REPL[200]:1

The same code runs fine in my coauthors Julia installation. I have tried to update, remove and reinstall the suspect packages (DataFrames, SharedArrays). Unfortunately I cannot simply uninstall Julia and perform a clean install, as this is happening in a HPC. Has anyone seen this before?

2

The same code runs fine in my coauthors Julia installation. I have tried to update, remove and reinstall the suspect packages (DataFrames, SharedArrays). Unfortunately I cannot simply uninstall Julia and perform a clean install, as this is happening in a HPC. Has anyone seen this before?

You should be using a Manifest.toml to control the versions of packages used in your project. Are you using the exact same versions of the packages your co-author is using?

3

Good point - thanks. Still, why would permutedims not work on a DataFrame?

4

Because it was only implemented in v0.22:

5

Thank you. This solved my issue. For some reason, DataFrames was only updating to 0.21.8 even after doing Pkg.update(DataFrames). I had to force installation of that version to get 0.22 installed.

6

Please use Manifest.toml files and proper package management rather than relying on updating the packages. Julia’s package management and compatibility system is really good!

1 Like
7

I got this error with DataFrames v1.3.6…

8

The exact error in the OP, a missing method for permutedims(::DataFrame, ::Int)? That seems unlikely, here’s DataFrames 1.3.6:

julia> df = string.(DataFrame(rand(5, 5), :auto));

julia> permutedims(df, 1)
4×6 DataFrame
(...)

or are you trying to call single arg permutedims?

julia> permutedims(df)
ERROR: MethodError: no method matching permutedims(::DataFrame)

for that you need to be on DataFrames 1.4 (latest release is 1.6, so there isn’t really a reason to use 1.3):

julia> permutedims(df)
5×5 DataFrame
(...)