Create the database, upload your data.
Import Pulsar and collect the data.
Pulsar structures your production data and merges it with offline data followed by the Pulsar metric analyzes your data.
Run the Pulsar analyzer on your notebook and get insights around your machine learning model.
Pulsar is model performance and management (MPM) software solution. This solution allow users to monitor their data inputs and model outputs and their evolution over time, across various drift and other statistical features. A core functionality of ‘alerts’ will allow automatic monitoring to relevant stakeholders when various performance metrics reach important thresholds.
Pulsar provides an easily integrated plug-n-play containerised service which provides different features like Data Ingestion, Metric calculations, Display feature metrics, Custom alert functionality. Data collected with the data-ingestion service can be used to calculate different metrics (data drift and performance) using user’s training (baseline) dataset for different time-intervals. Based on the same, Pulsar will use its scheduler service to monitor drift and trigger different standard and customised alerts.
Pulsar is being developed currently to augment different marketplaces (Azure, AWS) for the product to be available for users and be used along with other standard services to further enrichen the end-user experience.
Data Drift:
Peformance:
