DataVoyager

This package provides julia integration for the Voyager data exploration tool.

DataVoyager.jl can be used for data exploration. It can help you visualize and understand any data that is in a tabular format.

You can install the package at the Pkg REPL-mode with:

(v1.0) pkg> add DataVoyager

You create a new voyager window by calling Voyager:

using DataVoyager

v = Voyager()

By itself this is not very useful, the next step is to load some data into voyager. Lets assume your data is in a DataFrame:

using DataFrames, DataVoyager

data = DataFrame(a=rand(100), b=randn(100))

v = Voyager(data)

You can also use the pipe to load data into voyager:

using DataFrames, DataVoyager

data = DataFrame(a=rand(100), b=randn(100))

v = data |> Voyager()

With a more interesting data source

using VegaDatasets, DataVoyager

v = dataset("cars") |> Voyager()

You can load any IterableTables.jl source into voyager, i.e. not just DataFrames. For example, you can load some data from a CSV file with CSVFiles.jl, filter them with Query.jl and then visualize the result with voyager:

using FileIO, CSVFiles, Query, DataVoyager

v = load("data.csv") |> @filter(_.age>30) |> Voyager()

In this example the data is streamed directly into voyager and at no point is any DataFrame allocated.

You can also access a plot that you have created in the voyager UI from julia, for example to save the plot to disc.

You can access the currently active plot in a given voyager window v with the brackets syntax:

using VegaDatasets, DataVoyager, VegaLite

v = dataset("cars") |> Voyager()

plot1 = v[]

At this point plot1 will hold a standard VegaLite.jl plot object. You can use the normal VegaLite.jl functions to display such a plot, or save it to disc:

display(plot1)

plot1 |> save("figure1.pdf")

A useful pattern here is to save the plot as a vega-lite JSON file to disc, without the data:

using VegaDatasets, DataVoyager, VegaLite

v = dataset("cars") |> Voyager()

# Now create the plot in the UI

v[] |> save("figure1.vegalite")

At a later point you can then load this figure specification again, but pipe new data into it:

using VegaLite, VegaDatasets

dataset("cars") |> load("figure1.vegalite")