Package for working with tabular data in Julia using DataFrame's.

DataFrames.jl is now an installable package.

To install DataFrames.jl, use the following:

Pkg.add("DataFrames")

DataFrames.jl has one main module named DataFrames. You can load it as:

using DataFrames

• DataFrame for efficient tabular storage of two-dimensional data
• Minimized data copying
• Default columns can handle missing values (NA's) of any type
• PooledDataFrame for efficient storage of factor-like arrays for characters, integers, and other types
• Flexible indexing
• SubDataFrame for efficient subset referencing without copies
• Grouping operations inspired by plyr, pandas, and data.table
• Basic merge functionality
• stack and unstack for long/wide conversions
• Pipelining support (|) for many operations
• Several typical R-style functions, including head, tail, describe, unique, duplicated, with, within, and more
• Formula and design matrix implementation

Here's a minimal demo showing some grouping operations:

julia> using DataFrames

julia> d = DataFrame(quote     # expressions are one way to create a DataFrame
           x = randn(10)
           y = randn(10)
           i = rand(1:3,10)
           j = rand(1:3,10)
       end);

julia> dump(d)    # dump() is like R's str()
DataFrame  10 observations of 4 variables
  x: DataArray{Float64,1}(10) [-0.22496343871037897,-0.4033933555989207,0.6027847717547058,0.06671669747901597]
  y: DataArray{Float64,1}(10) [0.21904975091285417,-1.3275512477731726,2.266353546459277,-0.19840910239041679]
  i: DataArray{Int64,1}(10) [2,1,3,1]
  j: DataArray{Int64,1}(10) [3,2,1,2]

julia> head(d)
6x4 DataFrame:
                x         y i j
[1,]    -0.224963   0.21905 2 3
[2,]    -0.403393  -1.32755 1 2
[3,]     0.602785   2.26635 3 1
[4,]    0.0667167 -0.198409 1 2
[5,]      1.68303  -1.11183 1 3
[6,]     0.346034   1.68227 2 1

julia> d[1:3, ["x","y"]]     # indexing is similar to R's
3x2 DataFrame
                x        y
[1,]    -0.224963  0.21905
[2,]    -0.403393 -1.32755
[3,]     0.602785  2.26635

julia> # Group on column i, and pipe (|) that result to an expression
julia> # that creates the column x_sum.
julia> groupby(d, "i") | :(x_sum = sum(x))
3x2 DataFrame
        i    x_sum
[1,]    1  2.06822
[2,]    2 -1.80867
[3,]    3 0.319517

julia> groupby(d, "i") | :sum   # Another way to operate on a grouping
3x4 DataFrame
        i    x_sum    y_sum j_sum
[1,]    1  2.06822 -2.73985     8
[2,]    2 -1.80867  1.83489     7
[3,]    3 0.319517  1.03072     2

See demo/workflow_demo.jl for a basic demo of the parts of a Julian data workflow.

See demo/design_demo.jl for a more in-depth demo of DataFrame and related types and library.

• Library-style function reference
• Background and motivation

The Issues highlight a number of issues and ideas for enhancements. Here are some particular enhancements under way or under discussion:

• Data Streams: issue 34, doc

• Distributed DataFrames: issue 26

• Fast vector indexing: issue 24, commit

• Bitstypes with NA support: issue 45, doc

DataFrames fit well with Julia's syntax, but some features would improve the user experience, including keyword function arguments (Julia issue 485), "~" for easier expression syntax, and overloading "." for easier column access (df.colA). See here for a bit more information.

Please consider this a development preview. Many things work, but expect some rough edges. We hope that this can become a standard Julia package.