Create time series from object sets

To use Quiver’s time series transformations (such as derivatives or cumulative aggregates), your data must be in time series format. Data already modeled as time series in the Ontology can be found directly through the search bar under the Time Series tab.

When working in Quiver, it can also be useful to convert object sets with a timestamp property into time series. You can create time series from bar charts, line charts and categorical formula charts when (1) the x-axis is time and (2) only one data series is displayed.

To create a time series from an object set, follow these steps:

1. Add an object set that has a time property to your analysis.
2. In the next actions menu of the object set card, select Charts and then select either Bar chart or Line chart to add the chart. Alternatively, select the magnifying glass icon in the top menu bar, search for either the bar or line chart and then add the chosen chart to the analysis. Next, select the object set as input in the chart editor.

3. Group by a time property of the object. You can select the bucketing granularity you want; this will define the granularity of the time series created. In the example below, we chart the count of Stock events over time, by selecting Start Ts as the Group By property.

4. In the next actions menu of the chart, select Time series > Categorical plot time series.

A new time series chart will be created below the chart card. In this example, we converted the chart showing a count of Stock Events objects by time into a time series.

Any of the Quiver time series transformations can now be applied on top of this time series for further analysis. For example, to compute the cumulative count of events over time, we can add a cumulative aggregate on the time series plot.

Note that the time series derived from an object set is tied to this object set, and so will dynamically update based on changes to that input object set. For example, if you filter the input object set, the time series will change to only reflect the objects in that filtered object set. This allows for user workflows where time series can be generated dynamically based on various user selections.