Operational ML monitoring and a meter in one platform
- Jun 02
In real machine learning systems, it is important to conduct continuous monitoring of data and models. Even the ML model itself remained the same, the nature of the data could change, which could directly affect users. Today on the market there are many platforms designed for monitoring software where various system and business metrics are collected to reflect the most important data on visual dashboard and generate notifications. For example, Grafana, Datadog, Graphite, etc.
There are also tools for monitoring ML machine learning systems of the Neptune type, Amazon Sagemaker Model Monitor, Censius and other Mlops tools. But you can combine the observation of the machine learning system with classical engineering software monitoring on the same platform. This is achievable using the New Relic, the telemetry platform of remote monitoring of mobile and web applications, which allows you to collect, explore and receive warning about all telemetry data from any source in one place. Thanks to integration with many Open-Source tools, New Relic can work with various sources and receivers.
Sending data from ML systems to New Relic is implemented using the Python-PERFORMANCE-monitoring Python-Biblio, which is available on GitHub ().
Installation: ml_performance_monitoring
#machinelearning #artificialintelligence #ai #datascience #programming #Deechnology #Deeuplearning #bigData #bigdata