Create Visualizations

Visualization can help you explore your data and also share insights with others.

To Plot a dataset, you can use:

• Chart Builder
• Chat Box

tip

Before you get started, DataChat recommends:

• Using the Rename skill to make sure the columns you want to use in your visualizations have short, easy-to-read names.
• Making sure any time-based columns are placed on the X axis.

Chart Builder

The Chart Builder is an interactive interface that helps you create and customize in-depth visualizations.

Click the Plot button from the sidebar to open the Chart Builder.

The components of the Chart Builder:

• Chart Types
• Required Fields
• Optional Fields
• Dataset Sample Limit
• Filters
• Customize
• Chart Display
• Data Sample
• Describe
• Change Dataset

Chart Types

Select a chart type icon to build a chart. Hover over each icon in the DataChat app to view the chart types.

Required Fields

The required fields change to fit each chart type.

	X-axis	Y-axis	Aggregate	Bubble Size	Split	Density	Partition
Bar	✔️	✔️	✔️
Box	✔️	✔️
Bubble	✔️	✔️	✔️	✔️
Donut					✔️
Heatmap	✔️	✔️	✔️			✔️
Horizontal Bar	✔️	✔️	✔️
Line	✔️	✔️	✔️
Scatter	✔️	✔️	✔️
Stacked Area	✔️	✔️	✔️
Stacked Bar	✔️	✔️	✔️				✔️
Violin	✔️	✔️

• X-axis and Y-axis. The column whose values populate each axis.
• Aggregate. How the measured column (usually the y-axis) is aggregated. If no aggregate is specified:
  • For bar, heatmap, horizontal bar, line, and stacked bar charts, the average value is used.
  • For stacked area charts, the total value is used.
  • For all other charts, no default aggregate is used.
• Bubble size. The column whose values determine the size of each bubble.
• Split. The column whose values populate the donut chart.
• Density. The column whose values populate the cells of the heatmap.
• Partition. The column whose values divide the values of each bar into stacks.

Once you enter the required values, a preview of the new chart appears in the chart display. You can adjust values to view changes in real time before you click Submit and continue with your DataChat session.

Optional Fields

The optional fields change to fit each chart type and allow you to tune and detail your visualizations.

	Color	Group	Label	Overlay	Slider	Smooth	Split	Subplot	Slice Size	Slice Aggregate
Bar	✔️	✔️		✔️	✔️			✔️
Box	✔️				✔️			✔️
Bubble	✔️		✔️		✔️			✔️
Donut					✔️			✔️	✔️	✔️
Heatmap					✔️			✔️
Horizontal Bar	✔️	✔️		✔️	✔️			✔️
Line	✔️	✔️	✔️		✔️	✔️		✔️
Stacked Area	✔️	✔️			✔️			✔️
Scatter	✔️	✔️	✔️		✔️			✔️
Stacked Bar	✔️				✔️			✔️
Violin					✔️		✔️	✔️

• Color. The column whose values specify a color for each bubble.
• Group. The column whose values group each data point by each unique value.
• Label. The column whose values add a label above each point or bubble.
• Overlay. The column whose values populate an overlay (a second y-axis) on top of the existing chart.
• Slider. The column whose values create slider sections.
• Smooth. Determines whether the line is smoothed out from point to point.
• Split. The column whose values group the chart into sections. The column must include exactly two distinct values, such as "True" and "False".
• Subplot. The column whose values create separate plots for each unique value.
• Slice Size (donut charts only). The column values, grouped by the required Split column, to use to create the chart's slices. If this field is empty, the entire dataset is used and grouped by the required Split column.
• Slice Aggregate (donut charts only). The aggregation to use, if any, to aggregate the values in each slice.

Once you enter the optional fields, view the chart display to check your work. You can adjust values to view changes in real time before you click Submit and continue with your DataChat session.

Dataset Sample Limit

The Row Limit field in the Dataset Sample section allows you to specify the number of rows from your dataset to use as sample when plotting a chart. This helps to generate charts more quickly, especially when working with large datasets. By default, the Chart Builder automatically samples the first 15,000 rows.

Filter

You can use Filters to define expressions to limit the data shown in your chart.

Once you enter the filter parameters, click Apply. The labeled filter appears. Multiple filters can be added to a single chart. Saved filters can be removed by editing the chart in the Chart Builder and clicking the X on the labeled filter. You can also enable or disable a filter by clicking on it. The background of enabled filters is blue, while the background of disabled filters is gray.

note

Filtering with date or time types is currently not supported.

Customize

Change the appearance of your plots:

• Change labels, including the chart's title, axis labels (including the second Y-axis label when an overlay is used), and the caption.
• Use custom colors.
• Add annotations and captions.

Chart Display

Once you enter the required fields for the selected chart type, the chart display lets you view your changes in real time before you Submit the chart and continue with your DataChat session.

View Sample of Data

Click the arrow next to view sample of data to reveal a table that displays a small sample of the dataset.

note

If the chart is edited from an Insights Board, the data sample is unavailable.

Describe

Click the arrow next to describe to reveal a table that displays each column and its type. The table is a summary of Describe.

Change Dataset

Click the name of the dataset at the top of the Chart Builder to select a different dataset or a different version of a dataset. All Chart Builder fields except Chart Type are cleared.

Chat Box

Some chart types are only available using the chat box. Alongside the chart types available in the Chart Builder, the Plot skill also offers the following chart types:

• Confusion Matrix
• Correlation Matrix
• Geomap
• Hybrids
• Parallel Coordinates
• Ridgeline
• Sunburst
• Treemap

You can also use the chat box to create your visualization:

1. Enter Plot a and select the plot, chart, graph, or matrix you want to create.
2. Use the autocomplete suggestions to configure your axes, sliders, and any other options you need to complete your chart.

For example, to generate a scatter chart from the "Titanic" dataset that compares the age and fare of each passenger, grouped by gender, and sliding by class, enter in the chat box:

Plot a scatter chart with the x-axis Age, the y-axis Fare, for each Gender, and sliding by Pclass

The resulting scatter chart looks like this: