Objective

PredictiveWorks. supports tracking of data model artifacts and metadata. Metadata comprise model specific information such as name, life cycle stage or version, model parameters and model metrics.

Exploring which parameter selections caused which accuracy is an important step towards changing a model’s life cycle stage to ‘production’.

PredictiveWorks. offers a REST API to access and query its model registry. Model building results can be retrieved as time series data and enable visual model exploration.

Implementation

PredictiveWorks. model registry is exposed as a REST service, that returns query results in JSON format that directly supports Apache Echarts visualization widgets.

This is one of the easiest ways to enable companies to build their own data model dashboards.