Events Manager

See the definition of Events When used as labels for supervised or semi-supervised learning, they are specifically called Facts.

Components of Events

Purpose	Description
Learning Ranges	Time ranges to set boundaries of model learning. More detail here.
Labeling for Learning	Used as known conditions (facts) for training models. More detail here.
Evaluation Reference	Helps validate how well a model captures real conditions. More detail here.
Evaluation Ranges	Time ranges for bounded evaluations. More detail here.

Note

Events are Assessment Specific, unique to each Assessment within a Datastream.

Core Actions in Events Manager

Action	Description
View Events	Browse a list of events with fields like Start/End time, Event Name, Entity, Event Groups, and Duration.
Create Events	Use timeline gestures to draw events (Cmd/Ctrl + Click + Drag) directly on charts or create manually in the Events Manager.
Search Events	Filter events by name, group, entity, or batch ID. Advanced search allows for time-based filtering.
Delete Events	Select and remove one or multiple events.
Assign to Groups	Organize events into reusable Event Groups for learning and visualization.

Event Groups

Event Groups are labels used to organize multiple events under shared semantics like _normal, _failure, _bad_data.

Use Case Examples

Use Case	How It Helps
Model Training	Select event groups to define which events the model should learn from.
Visualization	Overlay event group data on charts to compare with model predictions.
Iteration	Quickly switch or update groups during model development.

Assigning Events to Groups

1. Select events from the list.

2. Click “Add to Groups”.

3. Choose existing groups or create a new one.

Best Practices

• Use Prefixes/Suffixes: Any group (Event group, Signal group, Entity group) name should start with an underscore "_" e.g., _normal, _supervised_failure

• Multiple Samples: Provide more than 3 examples of each condition. Better labeling has better results.

• Sub-categorize: Same event name across different groups if useful

• Separate Labels: Use different groups for learn, apply, and evaluate

• Multi-Labeling: Avoid overly generic labels that may encapsulate multiple different patterns (e.g., only “normal”)

Importing Events

Bulk import events via CSV or JSON, ideal for loading logs or maintenance records.

Import Steps:

1. Navigate to the Events tab.

2. Click the “Upload Events” (➕ icon).

3. Upload your file and fill in required mapping fields:

   • Start Time / End Time Column Name

   • Time Format (ISO 8601, Seconds, etc.)

   • Timezone (or default to datastream timezone)

   • Entity Column Name or static Entity

   • Event Name Column

   • (Optional) Event Group Column and Additional Tags

Sample Required CSV Columns:

Column	Description
start	Start time of the event
end	End time of the event
entity	Name of the entity
event	Name of the event label
event_group	Optional: Add in event group label

Exporting Events

To export event records for review or backup:

• Select events from the table.

• Click the “Export Events” button or menu option.

• A downloadable CSV file will appear in the Activities Monitor.