Tracking instance creation over time via computed timeseries

This configuration explains how to build a time series that displays the number of instances created over time (e.g. per month), even when the execution timestamp cannot be directly used in filters.

It introduces a two-step approach that enables accurate aggregation while remaining fully configurable.

Overview

The objective of this configuration is to:

• Track the number of instances created over time

• Overcome limitations related to filtering on execution timestamps

• Provide a reusable and scalable pattern for time-based metrics

• Allow retrospective analysis (works on already existing data)

Core principle

Since the execution timestamp (ExecutionInstant_i) cannot be directly used in filters for aggregation, this approach separates the logic into two steps:

1. Identify whether each instance was created during a given time period

2. Aggregate these results across all instances

Configuration

1. Instance-level time series (Boolean indicator)

A first time series is created at the instance level (e.g. Use Case).

This time series:

• Evaluates whether an instance was created during a given period

• Returns:

  • 1 if the instance was created during the period

  • 0 otherwise

• Acts as a building block for aggregation

Example formula (monthly)

The formula checks whether the instance creation date falls within the evaluated month

2. Aggregated time series (Global count)

A second time series is created to aggregate the results of the first one.

This time series:

• Uses the population of instances (e.g. all Use Cases)

• Applies the instance-level time series

• Sums all values

Since each instance contributes either 0 or 1, the result is the total number of instances created during each period.

Example formula (monthly)

3. Visualization

The aggregated time series can be displayed in a chart.

Flexibility

Although the example uses a monthly granularity, the same approach can be adapted to:

• Weekly

• Quarterly

• Yearly

• …

Key benefits

• Works despite filtering limitations on execution timestamps

• Fully reusable pattern

• Accurate aggregation across any population

• Enables retrospective analysis (no need for prior setup)

• Flexible time granularity

Summary

By combining:

• A per-instance time series (0/1 indicator)

• A global aggregation (sum)

This approach provides a simple and robust way to compute time-based metrics such as the number of instances created per period.