Imply provides a wide array of expressive visualizations that enable fast, interactive exploration of your data. Visualizations are not just static charts. You can interact directly with them by clicking on various graphical elements to select data segments for drill-down and deeper analysis.
When you change the shown dimensions, Pivot chooses the best visualization for the selected dimensions automatically.
To choose a visualization, click on the current visualization type to the right of the filter and split bars. The following example shows Table as the current visualization:
The following sections describe the available visualizations.
The default visualization is the Overall visualization. It presents a numeric summary of the currently selected measures, applying any selected filter criteria.
You can select multiple measures and show a time range comparison as well, as shown in the following screenshot:
If you create an Overall visualization with a single measure, you can apply conditional formatting to color the visualization as an indication of data severity—green for ok, amber for warning, red for critical.
To add conditional formatting, click the paintbrush icon in the visualization pane on the right, then click Add condition. Select the condition to apply to the data and the corresponding formatting. You can create multiple conditions.
You can also create a condition based on a comparison. In the following screenshot, the green color of the visualization indicates that the absolute change in number of events for the previous hour was less than 1,000::
If you add an Overall visualization tile with conditional formatting to a dashboard page, the colored icon next to the page name indicates the severity of the data on the page. See Dashboard pages for more information.
The bubble chart visualization lets you examine the relationship between two or three numeric variables. Each bubble in the chart corresponds to a single data point, and the measures for each point are indicated by the x-axis (the first measure), the y-axis (the second measure), and bubble size (an optional third measure).
The following example shows a bubble for each of the five channels in the filter. The bubble's horizontal position notes the number of events added to the channel during the latest six hours, and the vertical position notes the number of events deleted from the channel during the same period. The bubble size indicates the relative comment length for the channels.
The flat table visualization is similar to the table visualization but it displays flattened data instead of nested data.
Before you can use the flat table visualization, you must enable the SDK based visualizations feature flag with
"flat_table" in the array. For example,
["flat_table","gauge","time_series"] enables all SDK-based visualizations. See Feature flags for more information.
Select one or more Columns to Group data by and any additional Columns to display. You can select what to display when a column contains multiple values. Select a Pivot column and one or more Measures and the visualization displays a column for each combination.
The following example shows the Number of Events for City: Chicago in the Koalas to the Max data cube. The data is grouped by Agent category and Browser. The table shows the additional column Event Type. Where Event type contains multiple values, Pivot displays the latest value.
The gauge visualization displays a summary of a selected aggregate as a gauge.
Before you can use the gauge visualization, you must enable the SDK based visualizations feature flag with
"gauge" in the array. For example,
["flat_table","gauge","time_series"] enables all SDK-based visualizations. See Feature flags for more information.
The gauge shows a number or percentage proportional to the Min and Max values you set. You can color specified ranges, set a custom label, and show a legend.
The following example shows the number of events in the Koalas to the Max data cube, proportional to the maximum 350,000. The legend shows three numeric ranges colored green, orange, and red.
The geo marks visualization is the natural choice for dimensions that represent geographically encoded data. It can work with country encoded data. To use geo-oriented visualizations, modify the data cube configuration to make the country encoded data to be Geo type.
The geo shade visualization (also known as a Choropleth map) is another choice for dimensions that represent geographically encoded data. It can work with country encoded data. To use geo-oriented visualizations, modify the data cube configuration to make the country encoded data to be Geo type.
The heatmap is a visualization that shows two dimensions as a 'matrix'. It is similar to the grid visualization and it works particularly well when one or both of the dimensions are continuous.
The horizontal bars visualization shows each dimension in horizontally-oriented time buckets.
The line chart visualization is designed to demonstrate a trend over time. It's useful for showing temporal dimensions.
The following example line chart displays two measures—Number of Events and Session Length—over time for four Country values.
You can display two continuous metrics on the same chart with two axes. To do this, select two measures and then click the paintbrush icon in the right pane. Select Show measures in: Cell and Dual axis: Yes.
The following example line chart displays the same Number of events and Session length measures over time as the above chart. On the dual axis chart, the data for the four selected countries displays as a single line for each measure.
You can use a Markdown tile to add formatted text to a dashboard. For more information, see Managing dashboards.
The records visualization shows the raw data underlying the data cube, allowing you to see all dimensions that are in each record. This view can be useful for certain debugging issues. You can also view your records in groups for specific values of one or more dimensions by adding the dimensions to the show bar.
The records table visualization shows the raw data underlying the data cube, in table format. To filter the data, click any value in the table and select Exclude.
The Sankey visualization shows the flow from one set of values to another. In Sankey diagrams, the width of each connection is proportional to the flow volume between the values.
In contrast to the line chart visualization that displays lines for multiple dimensions on a single chart, the sparkline visualization displays multiple line charts—one for each dimension.
The following example sparkline chart displays the number of events over time for five Country Name values:
Hover over the first column heading—Country Name in the above example—to display the Swap axes icon. Click the icon to swap the splits assigned to the axes.
The spot matrix shows the distribution of events on a two dimensional matrix using circles, with larger circles representing a greater number of events.
The stacked area chart is similar to the line chart, and can be used in similar situations. Unlike the line chart, the area chart lets you see what all the values add up to together. This lets you see the overall trend, while also showing the individual contributions of different dimensions.
If your event data contains latitude and longitude coordinates, you can use the street map visualization to pinpoint events to precise locations on a map.
If your Imply deployment is on-prem, contact your Imply support representative to enable the street map visualization.
Create dimensions for your latitude and longitude data. Add the dimensions to the show bar and select Street map from the visualization selector on the right side of the page.
You can click the paintbrush icon to change the default Grid display to Blobs, for a circular representation of the data. The darker the grid or blob color, the higher the concentration of events at that location. Click anywhere in a grid or blob to see the corresponding latitude and longitude coordinates and the number of related events. Click Filter to filter the event display by specific latitude and longitude coordinates.
The sunburst (also known as a donut or pie chart when using only one split) is a visualization that represents the ratios between the values of a dimension. When rendering multiple dimensions, each is subdivided to show proportional representation.
The table visualization is the default view for a categorical (string) dimension. It presents a table view of the data with formatting that aids visualization of measure magnitude.
The table visualization can show multiple dimensions, as well as multiple measures, as columns.
The following example shows the number of Wikipedia events for City Name
New York by Channel for the latest 6 hours, with a comparison to the previous period:
Click the paintbrush icon on the right to access layout options. Polaris displays a nested layout by default—shown above. Click and drag the dimensions in the show bar to rearrange their order and change the nesting.
Hover over the first column heading—City Name > Channel in the above example—to display the Swap axes icon. Click the icon to swap the splits assigned to the axes.
Select the Flat row layout to display a column for each dimension.
The Overall rows display the dimension totals. In the above example, there are 10 events that match the applied filters—9 for the
#en.wikipedia Channel and 1 for
#pl.wikipedia. Of the 9
#en.wikipedia events, 5 are for City Name
London and 4 are for
The time series visualization allows you to use time series functions to generate a line or bar chart showing the rate of change in your data.
Before you can use the time series visualization, you must:
- Load the
imply-timeseriesextension. See Time series functions: Setup for more information.
- Enable the SDK based visualizations feature flag with
"time_series"in the array. For example,
["flat_table","gauge","time_series"]enables all SDK-based visualizations. See Feature flags for more information.
In a data cube, select Time series from the drop-down list of visualizations in the top left corner, then set the following properties:
- Timeseries function: Select TIMESERIES to create a time series of the data points, or DELTA_TIMESERIES to create a time series that contains delta differences between the data points. See Time series functions for more information about each function.
- Timeseries column: The column Pivot applies the time series to—it must be numeric. The time series column displays on the y axis. The default time column for the data cube displays on the x axis.
- Group by: The column to represent as lines or bars in the visualization.
- Sorted by: The measure by which to sort the data.
- Group limit: A limit to apply to the group column.
You can refine the visualization by customizing the additional properties:
- Pre-bucketing: Shift all data points to the nearest time period selected.
- Timeseries bucket: If you select the DELTA_TIMESERIES function, choose an interval for which the deltas are added up (or folded) within a time bucket.
- Time-weighted average: Transform the time series data points to the average value of the Timeseries column, within the selected time range.
- Interpolator: If you enable Time-weighted average, you can specify the method to interpolate missing data points in the time series:
- Padding: Carry forward the closest value in the series.
- Backfill: Carry backward the closest value in the series.
- Linear: Use linear interpolation to fill the missing data points.
- Limit: The maximum number of data points in the time series.
If Pivot displays the error
Exceeded the max entries allowed, try increasing the Limit by x10. For example, if the current limit is 10,000 change it to 100,000.
The following example visualization applies the TIMESERIES function to a Wikipedia data cube. The time series column is Added and the data is grouped by Channel. The visualization displays the top four channels with the largest Number of Events. The Time weighted average is set to Hour—there is a point on the chart for every hour as shown on the x axis.
The treemap allows you to see how the values of a dimension combine to make up the whole. It is particularly suitable for hierarchical dimensions.
The vertical bars visualization shows each dimension in vertically-oriented time buckets.