One of the most powerful ways to share results of your data analysis is through visualizations. On data.world you have access to both integrations for many third-party tools and also to Chart Builder, a visual editor for Vega-Lite built specifically for data.world. It is the perfect tool for those looking to create a simple visualization that is lightweight and easy to embed.
Visualizations in Chart Builder can be made from either a file or a query. They can be saved in various formats, shared, or embedded in the various Simple Editor windows.
In this article we'll explore:
- Using Chart Builder with files
- Using Chart Builder with queries
- Formatting options for visualizations
- Charting more than one column in graph
- Sharing and saving visualizations
You can create visualizations of tabular data files in either projects or datasets without ever writing a query. The dataset Shark attack data updated daily and the GSAF5.xls.xlsx are used in this example. To create a visualization of all the data in the file, select the Explore icon to the right of the file name on the overview page:
You'll be taken to the data workspace and a view of all the data in the file and you can build a chart of it by selecting the Open in app button:
From there you can choose the application you would like to use to create your visualization. In this exercise we'll use Chart Builder. Chart builder comes with two options for creating and modifying charts: a Visual Builder and a Vega-Lite Editor. The easiest way to use it is to create your initial chart on the Visual Builder tab and then switch over to the Vega-Lite Editor to make any changes outside the scope of the Visual Builder. Detailed information on how to use Vega-Lite to customize your charts is available on the Vega website.
To create a quick bar chart of the number of people in the dataset who have been fatally attacked by sharks, select the field fatal_y_n from the dropdown list for the X axis, and COUNT(*) from the bottom of the list on the Y axis. Like magic, our chart appears on the right side of the screen:
Chart Builder can also be used on queries. Using a query as the basis for your chart enables you to:
- Clean up data (e.g., remove NULL values)
- Filter out data (e.g., specify a time period, a specific value, an aggregation, etc.)
- Change your data structure so that it can be charted
In the shark attack dataset referenced above there is a saved query called Query for analysis by year or country This query has been written to exclude NULL values and remove non-binary entries on specific fields. It can be used to create a number of different charts:
Click the Open in app button above the results to get to Chart Builder.
Continuing with the example above, to make a chart with circles for marks that compares the number of attacks on men vs on women across the years select Circle from the Marks dropdown, and year from the X axis dropdown:
If the axis doesn't display the way you want it to, you can override the default format for the Type under the Options dropdown. In this case the year was read as a number because of the underlying data type and the type was set to Quantitative when Ordinal was the right choice:
Set the Y axis to COUNT(*) and Color to sex:
If your chart is not appropriately sized for viewing you can manually set the chart size. A size that shows the data of this chart to best effect is a width of 950 and a height of 730:
Another handy thing you can do in the options section is order the results in your graph. You may have ordered them already in your query results, but that order does not carry over to the graph. For this example we'll use the saved query Countries with >10 unprovoked attacks since 1960 with mortality data. After running the query, click Chart, set the X axis to country and the Y axis to # attacks:
To sort by the country with he most attacks select Options on the X axis, choose Descending for the sort and y - # attacks on the field:
To add information on if the attacks were fatal or not, select fatal_y_n next to color.
If your results do not seem to display as they should, check to make sure the field you are sorting on is not being improperly aggregated. Being able to switch over to the Vega-Lite Editor is very handy for identifying this kind of configuration issues. Looking in the Vega-Lite Editor in the area dealing with "sort" the operation is set to "average":
In this case "sum" is the correct option, and upon replacing "average" with "sum" the visualization displays properly:
NOTE: Once you have made edits in the Vega-Lite Editor you can no longer make any changes in the Visual Builder so save your Vega-Lite Editor changes for when you are finished building the chart with the Visual Builder.
It is occasionally the case that you'll want to include the data from more than one column in a chart. In some cases the values in two columns are related and could have been stored in one column but weren't, e.g., separate columns for Male/Female, high-temp/low-temp, max value/min value, start date/end date, etc. In other cases the data in the columns isn't related so you wouldn't expect it to be stored in the same column, but you still want to use it all together in the same graph. It is possible to view the data from multiple columns in Chart Builder, it just requires modifying your original query with UNPIVOT so that all the columns you would like to view are combined into one column in your query results. Detailed instructions can be found in our article Using multiple columns of data in Chart Builder visualizations.
Chart-builder visualizations on data.world can be saved in a number of json, image, and html formats shown under the Download button:
There are also a variety of options for sharing your visualization on data.world:
Selecting Share > Insight lets you add your visualization to any project on data.world for which you have permission:
To share the insight you are prompted to chose a project where you will share it, to give it a title, and optionally to add comments. The final option (selected by default) is to save the visualization as a Vega-Lite source file on the project:
Share > File lets you add the visualization to any dataset or project for which you have permission. Share > Markdown Embed (Comment) you can embed your chart in any place which uses Markdown (e.g., insights, comments, summaries). By default the embedded chart will be a static rendering of the data from when the visualization was created. However using the Vega-Lite Editor you can create a 'live' chart that updates as the data on which it's based updates. The shark attack dataset is an example of continually-updating data.
To make a chart 'live', go to the Vega-Lite Editor and scroll down to the section referencing the "data" parameters:
Under the "data" element replace "source" with "url" and add a hardcoded url for the query that drives the visualization (you get this in the workspace while viewing the query), and add a "format" element with the type "csv":
Then when you select Share > Markdown Embed (Comment), you'll get Markdown text for a live version of the visualization that you can copy and paste into Insights, Comments, and Summaries on data.world:
Here is an example of the live visualization above used as an insight on a project that uses the dataset:
Finally, if you want to share a link to the Chart Builder screen for the visualization so someone else can edit and run it, you can do so with the Share > URL option:
This example can also be found on the discussions page for the dataset. Notice the use of Markdown stying to make the comment more readable. Detailed instructions for using Markdown on data.world can be found here.