Updating Events

Fiddler supports partial updates of events. Specifically, for your target column and metadata columns.

The most common use case for this functionality is updating ground truth labels. Existing events that lack ground truth labels can be updated once the actual values are discovered. Or you might find that the initially uploaded labels are wrong and wish to correct them.

We also support updating metadata columns with new values. You can observe the changes in Analytics, but the Performance metrics of the columns won't be reflected.

Other columns(Decision, Input and Output) can only be sent at insertion time (with update_event=False). They are dropped if you send them and set update_event=True.

Set update_event=True to indicate that you are updating an existing event. You only need to provide the fields that you want to change—any fields you leave out will remain as they were before the update.

---

Example in client 3.x

Below is the code you can use to update the target and metadata columns values for selected events (for instance when you get ground truth data or some other information relating to the event at a later stage)

model_spec = fdl.ModelSpec(
    inputs=['input1'],
    targets=['target']
    outputs='output'
    metadata=['uri', 'record_tag'],
)

fdl_model = fdl.Model.from_data(
    name=model_name,
    project_id=project.id,
    source=dataset_df,
    spec=model_spec,
    event_id_col='event_id'
)

model.publish(source='my_events.csv', update=True)

Where my_events.csv contains the corresponding columns to be updated, as well as the event_id_col of the model: event_id. Here is an example of my_events.csv:

event_id	target	uri	record_tag
A1	Selected	s3://dataset/image.jpg	category_1

Refer to publish doc for more details on different sources and parameters.

\

📘 There are a few points to be aware of:

• Performance metrics (available from the Performance tab of the Monitor page) will be computed as events are updated.

  • For example, if the ground truth values are originally missing from events in a given time range, there will be no performance metrics available for that time range. Once the events are updated, performance metrics will be computed and will populate the monitoring charts.

  • Events that do not originally have ground truth labels should be uploaded with empty values—not dummy values. If dummy values are used, you will have improper performance metrics, and once the new values come in, the old, incorrect values will still be present.

  • Update on Metadata columns won't be reflected.

• In order to update existing events, you will need access to the event IDs used at the time of upload. If you do not have access to those event IDs, you can find them by using the fdl.Model.get_slice API and checking the __event_id column from the resulting DataFrame.

• If you pass an updated timestamp for an existing event, this timestamp will be used for plotting decisions and computed performance metrics on the Monitor page. That is, the bin for which data will appear will depend on the new timestamp, not the old one.