Visualize

The visualize tab provides all the machinery for

1. performing preliminary data exploration;

2. visualizing the output of the computational tabs (prediction, clustering, dimensionality reduction).

Fields

The tab consists of two main fields: Attributes and Layers. The way simple analyses are performed and rendered follows the philosophy of AlgebraOfGraphics.jl.

• Attributes. The attributes field contains the variable to be plotted and additional attributes that will be applied globally to the plot to be rendered. For instance, the syntax :column_1 :column2 implies that the two selected columns will be used as x and y axis of any plot that will be produced from such attributes. Adding more attributes such as :column_1 :column2 color:column_3 means that the column_3 (being continuous or categorical) will be used to establish the color code associated with rendered plot. See examples in the following sections.
• Layers. The layer field can be filled with one or more operations that will take the content of the Attributes filed as input (and specifications) and result in a series of superposed plots. For instance the syntax: Attributes:column_1 :column2 color:column_3 Layers: scatter produces a scatter plot having column_1 and column_2 as x and y axis, respectively, and whose points are colored according to column_3. The + operator can be used to combine layers. Thus, the syntax Attributes:column_1 :column2 color:column_3 Layers: scatter + linear renders a scatter plot and a plots linear models according to the grouping defined by column_3. Specifications relative to a specific operations can be added directly in the Layers field. See the sections below for specific examples.

In the following sections, we will use the palmerpenguins dataset to describe the most common computational steps in data exploration.

Layers

Frequency

Dodge

Stack

Scatter

Color

Marker

Linear

Density

Contour

Combine layers

Split by categorical variable