Event set cards

Back to: Index of cards

Cards in this section are used to visualize and analyze events in bulk. An event consists of a start and end timestamp, similar to a time range, but can also be enriched with other data to support analysis.

• Events plots
• Event indicator series
• Event set from ranges
• Event set from tabular data
• Linked event set
• Time series search
• Time shift event set
• Event comparison plot
• Reference profile bounds

Events plots

Every event set is visualized using an Events plot, which plots events on a time axis at the bottom of the chart.

• Hovering over each event will show the details of that event. You can configure what details to display for each event through the Tooltip settings.
• Events are grouped based on the zoom level of the chart. Hovering over an event group shows a list of events in that group.
• The input data must have a date or timestamp property or column for it to be plottable as multiple events.

Event sets and events plots are backed by transform tables, where each row is treated as a single event. The backing transform table can be accessed through the next actions menu of the event set by selecting Convert > New transform table. You can use the backing transform table to modify or enrich the underlying event data. For example, you may want to compute a color label column for each event. You can then use the column to color code in the Colors settings under Event Options in the Display tab of the Editor panel.

Event indicator series

An event indicator series creates a time series out of an event set indicating the number of events occurring at a given time.

• Note that only events with non-zero durations will be plotted.
• To manually shift the start or end time of the events, use the time shift event set card to modify the input event set.

Event set from ranges

Create an event set by specifying a list of time ranges for each event. Create each event by selecting a range parameter, or separately inputting start and end timestamps. Customize the title and description that appear in an event's tooltip by editing the corresponding fields in the Input event data section in the Data tab of the editor panel. When a range parameter is used, its title and annotation automatically populate the event's title and description, if available.

Event set from tabular data

Create an event set from a tabular data source such as an object set, transform table, or materialization. The start and end timestamps of each event are populated using a column or property from the input table, specified in the Events configuration settings in the Data tab of the editor panel.

Linked event set

Create an event set by starting from a base object and finding related objects through Ontology-defined links (also referred to as a search around). Multiple layers of object relations can be traversed to build an event set of the desired type, and then the event set can optionally be filtered. Set the start and end timestamps for the events by specifying which object properties hold this data in the Events configuration settings in the Data tab of the editor panel. A linked event set can also be created from a starting time series. In this case, the root object of the time series will be used as the base object. Linked event sets are object-based, so the event tooltips are populated using data from each object.

Time series search

Visually display when certain conditions are met on one or more time series plots, such as when the source series exceeds value 100 and another series hits a specified threshold. The time series search formula should return a Boolean value, and an event will be created over every time interval in which this Boolean value is true. You can use Automate to save the identified events as objects in the Ontology.

• Time series search results are displayed as events in an events plot.
• Run a single search or multiple searches:
  • Single: Performs a search on a single time series plot, returning one event for each time interval that satisfies the specified conditions.
  • Multi: Performs a search across each row of a transform table (limited to 1,000 rows) that returns one event for each time interval that satisfies the specified conditions. Search properties in the table using @property syntax.
• Optionally set multiple search conditions by selecting + next to Apply. Values that match all conditions will be searched for by default. Change the search to match any condition by switching the Join type from AND to OR.
• Optionally set a minimum or maximum condition to find formulas that are true for a specific duration of time. For example, find only the events that were longer than one hour, or shorter than one day.
• Reference any time series plot or parameters in your analysis when writing the formula in the Conditions text box (for example, $AN < 5 && $W = $X).

Time shift event set

Modify an existing event set by shifting the start and end timestamps forwards or backwards by a duration.

Event comparison plot

Compare the behavior of a time series across multiple events by isolating and aligning the segments of the time series where each event is occurring. The plot displays the time series segments in relative time, so that the value at the each event's start time is aligned to zero. The image below shows how the event comparison plot can be used to visualize the behavior of temperature in a tea vat over 80 degrees.

Reference profile bounds

Create a bounded region representing the variability among multiple time series, which is useful for monitoring other time series against these bounds. The bounds at each timestamp are computed by performing the following steps:

1. Aggregate the points at that time stamp to calculate the average and standard deviation.
2. Generate the upper bound by summing the average and standard deviation.
3. Generate the lower bound by subtracting the standard deviation from the average.

Use relative time series to ensure that there is overlapping data to be aggregated. Relative time series groups can be created using the event comparison plot, or by using the relative time series transform on an existing time series group. The example below shows how reference profile bounds can be created from the humidity data of high-quality tea batches and used to monitor humidity for other batches.