In a Voronoi treemap, you can see the values of the columns you add, portrayed as a tessellation of polygons whose proportions depend on a numeric column you choose. These polygons may be subdivided into smaller polygons and constitute a hierarchical structure with as many levels as those into which the data is divided (there is a legend above the chart explaining the hierarchy of data).
What data do I need for this widget?
The option to create this chart will be disabled unless your query contains at least two columns, one of them with numeric values. Furthermore, to show meaningful content on the chart, you must group your data by at least two keys using a no-time option. Also, it is highly advisable to add some aggregation functions to provide mathematical significance to the variables you want to analyze.
If you grouped using a time option, the diagram will only show the data for the latest period available for the time range specified.
Creating a Voronoi treemap
Here we describe how to create this chart using examples. We want to analyze the distribution of flights and flight delays depending on the airline.
Let's go step by step through the process with another example:
Go to Data Search and open the required table.
Perform the required operations to get the data you want to use in the chart.
- Click the gear icon on the toolbar and select Charts → Diagram → Voronoi treemap.
Click and drag the column headers to the corresponding fields.
Required field Description Data type Signals The variables to analyze, whose unique values will be represented as separate polygons in the diagram and thus will determine the number of polygons into which the diagram will be divided. You can add as many as you need to construct a hierarchy of subdivisions, which will be determined by the order of addition, the first corresponding to the top level and the last to the bottom level. Any type but json or bigint Value The measuring variables whose values will be used to establish the proportions of each of the polygons. You can add more than one in order to have several measuring criteria but they cannot be used at the same time. You can switch between them inside the diagram but the first you add will be used by default. float, integer, duration
The Voronoi treemap is displayed. This is a visual depiction of the average response length of the connections to Devo in each city and with each response time over one day.
Customizing the Voronoi treemap
Several options for customizing how you want to visualize this information appear above the treemap:
|Size by||This dropdown contains those variables added to the value field when generating the treemap. You can switch between them to use their values as the criteria to establish the proportions of the cells.|
In this example, you can choose between determining cell size based on the average response length (avg_responseLength) for each grouping occurrence (city-responseTime) or the number of events for each grouping occurrence (Count).
|Color by||By default, cell colors are applied randomly without any color spectrum that visually correlates cells and values. This dropdown contains those variables added to the value field when generating the treemap and selecting one of them will establish a color spectrum using its minimum and maximum values as cardinal points (green-red). Cells will be colored to reflect the position of their value inside that spectrum.|
|Partitioning||By default, cell outlines are colored in black without using a pattern that visually correlates outlines and values. This dropdown contains those variables added to the value field when generating the treemap and selecting one of them will color the cell outlines to reflect the status of their values in relation to the optimal status (green to red).|
|Search||Enter text to search for a value in the Voronoi map. The cells containing the specified string will be highlighted, together with their children if any.|
|Filter||Enter text to filter the chart by a specific value. Only the cells matching the string will be shown, together with their children if any.|
|Legend||This is a path that lists the fields added as signals. The first signal in the path is the primary cell grouping, followed by subsequent groupings. Click and drag to change the order of the signals in the path to change the grouping order for cells in the chart.|
|Values||When drilling-down into the cells, this shows the values of each grouping.|
Click the information icon to display a list of keyboard shortcuts that you will use to navigate the chart and to modify its style and layout:
Types of visualization
Cells are drawn from left to right according to their weight (from larger to smaller).
Larger cells appear in the center.
Smaller cells appear in the center.
Cells are randomly distributed.
Cells are drawn from top to bottom according to their weight (from larger to smaller). A cell's size corresponds to its weight.
S Squarified Treemap
Cells are rectangular and drawn from left to right according to their weight (from larger to smaller). A cell's size corresponds to its weight.
D Squarified Ordered
Cells are rectangular and drawn from top to bottom according to their weight (from larger to smaller). A cell's size corresponds to its weight.
P Flattened / Not Flattened visualization
Flattened visualization of all branches, without having to do a drill-down. The Flattened option can be applied to all previous visualization options. Pressing P again will return it to the normal visualization.
Options applicable to all visualizations
Select group, click again to deselect.
CTRL + Left click
Select multiple groups.
Left double click
SHIFT + Left double click
Focus on a group. Select a group of cells and analyze them as if they were a separate Voronoi map.
Open a group of cells (drill-down).
Right double click
SHIFT + Right double click
Zoom in / out
Pan around zoomed visualization.
Unexpose & close all groups.
Displays detailed information in each cell. Left-click on the cell and then press G. A panel will open to the right to display the following information:
Compare multiple cells:
Show/hide percentage in the legend.
Show/hide value in the legend.
Show/hide the percentage of the total in the legend.
Show/hide zero values.
Calculate cell proportions using a logarithmic value to normalize size. This way, nodes with 20% of the total weight can be the same size as nodes with 5% of the total weight. This option is recommended when having very polarized values and needing an exhaustive approach in which value omission is not admissible for the purpose of the analysis. This way, we avoid having very tiny cells that might be almost invisible or even omitted from the diagram.
Calculate cell proportions using real aggregated values so a cell's size corresponds to its weight. Nodes with 20% of the total weight will represent 20% of map's total size, whereas nodes with 1% of the total weight may not even be seen. This option is recommended when needing a generalistic approach in which marginal values can be omitted without substantially affecting the analysis. This way the correspondence of values and sizes can be easily identified at a glance.
Show/hide values as bytes.
Show/hide values as time (Chrono style).
Show/hide this help.
Other Voronoi layouts
Visualization of the average response length (avg_responseLength column) per city over the last day and comparison of three of the cities.
- Press G and select Knoxville and Alcorcon by pressing CTRL + clicking the cells. Move the mouse over the Madrid cell to compare it to the other ones.
The cells selected will appear at the lower part of the right panel while the cell over which you hover will appear at the top, showing in both cases their information and aggregated values. The one at the top will be compared to those at the bottom, showing their differences in red or green (fewer or worse).
- Hit the G key again when you finish to go back and remove the comparison panel to the right.
Visualization of the average response length (avg_responseLength column) per city over the last day, ordered by size and colored using a spectrum based on the average response length (avg_responseLength column).
- Press D to apply the Squarified Ordered visualization.
- Press N to calculate cell size using their weight.
- Select avg_responseLength in the Color by field.
The spectrum of colors shows the average response length (maximum is red, minimum is green) and the cell size represents also the average response length (the larger they are, the longer the response). With this visualization, we establish a correlation between size and color, making it easier to spot potential problems by looking at big red cells.
You can recreate the example explained above with the data from the following query and mapping the fields as follows:
from siem.logtrust.web.activity group by city, responseTime every - select avg(responseLength) as avg_responseLength, count() as count
|Required field||Column added|
The following video shows how to create and use a Voronoi chart to analyze your data. It does so by comparing the two different versions available in Devo (Data Search and Activeboards) so that you can choose the best for you.