Events & Batches
Data that belongs to a batch, a run or a cycle rather than to a moment.
Time-series answers what was the temperature at 14:05. A lot of industrial questions are not shaped like that: what was the average temperature during batch 4471, which orders ran on this line last week, was the mixer in cleaning or in production at the time.
Events attach that identity to your data — a name or an ID covering a span of time — so you can query by the thing that was happening instead of only by the clock. They are useful whenever the meaningful unit is a batch, a production order, a cycle, a shift or a machine state.
Where to start
| If you want to | Read |
|---|---|
| Work out how to model your batches, orders or states — the group system, the datapoints and how they attach to twins | Design your Event/Batch solution |
| Get event data into the platform | Learn how to upload event data |
| Handle a twin that is in more than one state at once | Working with simultaneous event statuses |
| Query it, on its own or alongside time-series | Querying Event datapoints |
Design first. An event solution is mostly modelling: once data is uploaded against a group system, changing your mind about the structure means reworking what is already stored.
In pipelines
Event datapoints are queryable from a Query step like any other datapoint, which is what makes them usable as pipeline inputs — training a model per product grade, or comparing a cycle against the previous one, rather than against an arbitrary time window.
Updated 25 days ago