What is Graphsignal?

Graphsignal is a machine learning model monitoring platform. It helps ML engineers and data scientists address data issues and analyze model performance in production.

How it works

When machine learning models are deployed to production, their input and output data should be monitored. Graphsignal loggers provide a simple API to log data. The logger automatically windows the data, computes statistics and sends it to Graphsignal platform for further processing.

It is essential to monitor model performance metrics such as accuracy for production ML models. Graphsignal provides a simple way to log ground truth labels, automatically computes and sends performance metrics for dashboarding and issue detection.

The Graphsignal platform processes the statistics and other logged data further to store it and detect issues. The data statistics, alerts and metrics are then available via dashboards at app.graphsignal.com.

Graphsignal diagram

Getting started

  • Sign up for an account.
  • See Python Quick Start guide on how to add logger to your ML application, pipeline or batch jobs.