Data tag messages

Insights & Data Tags¶

In the Kelvin UI, Data Tags are now known as Insights.

Data Tags may be depreciated in the Kelvin SDK and Kelvin API in future releases.

Keep an eye on future releases for updated announcements.

Insights Management¶

You can also manage your insights using the Kelvin SDK and Kelvin API.

You can read all the details in the Platform Administration in the Insights section.

Data Tag Messages¶

Insights are used to label Asset time series data. This has many applications such as;

• For data scientists to use for machine learning training
• For developers to have additional information beyond the raw data to review and improve Kelvin SmartApps™
• For Operations to tag events linked to the time series data for historical records

On the Kelvin UI, the insights will show in the Data Explorer.

Kelvin SmartApps™ can publish a DataTag Message in order to label important events on a certain point in time (or time range) for a certain Asset.

The DataTag Object supports the following attributes:

app.yaml setup¶

To publish insights, declare the tag names your SmartApp will produce in the data_tags.outputs section of your app.yaml. Each entry is the tag name as a string:

1
2
3
4

data_tags:
  outputs:
    - anomaly_window
    - my_tag

Publish a Data Tag¶

This is how they can be created and published:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

from datetime import timedelta, datetime

from kelvin.application import KelvinApp
from kelvin.message import DataTag
from kelvin.krn import KRNAsset, KRNDatastream

app = KelvinApp()

DATATAG_DURATION_SECS = 60

@app.timer(interval=10)
async def publish_data():

    now = datetime.now().astimezone()

    # Create and Publish DataTag
    await app.publish(
        DataTag(
            start_date=now - timedelta(seconds=DATATAG_DURATION_SECS),
            end_date=now,
            tag_name="my_tag",
            resource=KRNAsset("my_asset"),
            contexts=[KRNDatastream("my_datastream")],
        )
    )

app.run()