🤖 A(I)PI app

Overview of 24SEA's API endpoints for Artificial Intelligence models predictions.

Fetch model metadata from the registry via SQL-like queries.

These endpoints can fetch one or more models from the registry and make a prediction based on the input data provided.
The models normally work with aggregate data, as they naturally accept input in the format of 24SEA Datasignals API output, where aggregations are generally made on a 10-minute basis, although this is not a strict requirement. Therefore, the input data is expected to always contain a column named timestamp, passed as ISO8601-compatible string. The input data may contain additional columns, but the 'timestamp', while <value> columns change depending on the model that will be invoked.

To date, the following sets of endpoints are available:

Model Validity Model Used Input-data Output
Location Specific Timestamp-compatible model 10-min aggregates Prediction
Location Specific Timestamp-compatible model 10-min aggregates Schema
Location Specific Version-specific model 10-min aggregates Prediction
Location Specific Version-specific model 10-min aggregates Schema
Fleetwide Timestamp-compatible model 10-min aggregates Prediction
Fleetwide Timestamp-compatible model 10-min aggregates Schema
Fleetwide Version-specific model 10-min aggregates Prediction
Fleetwide Version-specific model 10-min aggregates Schema
🔎 How is a location-specific model identified
<statistic>_<site_id>_<location_id>_<short_hand>


🔎 How is a fleetwide model identified
<statistic>_<site_id>_<short_hand>_pred


The Prediction endpoints return a dictionary containing the model name, version, start validity date, training date, and the prediction for the model. If the prediction is not a scalar, the prediction is returned as a dictionary.

The Schema endpoints return a dictionary containing the model name, version, start validity date, training date, and input schema and output schemas. If the schema changes between different model versions, the input data passed to the model must be compatible with the union of all input schemas.

This is an example README.md file inserted in a YAML file.
This is a sub-app and here you see an example table.

List Minimum Maximum Average
[1, 2, 3] 1 3 2
[4, 5, 6] 4 6 5