Skip to content

Latest commit

 

History

History
 
 

algorithmia

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
img
img
 
 
algorithmia.yml
algorithmia.yml
 
 
readme.md
readme.md
 
 

readme.md

Algorithmia

Provided by: Algorithmia

Algorithmia is an MLOps platform that includes capabilities for data scientists, application developers, and IT operators to deploy, manage, govern, and secure machine learning and other probabilistic models in production.

Algorithmia Insights is a feature of Algorithmia Enterprise and provides a metrics pipeline that can be used to instrument, measure, and monitor your machine learning models.

Algorithmia Model Risk Dashboard

This template allows you to stream operational metrics and user-defined, inference-related metrics from Algorithmia to InfluxDB using Telegraf and Kafka.

Quick Install

InfluxDB UI

In the InfluxDB UI, go to Settings->Templates and enter this URL: https://raw.githubusercontent.com/influxdata/community-templates/master/algorithmia/algorithmia.yml

Influx CLI

If you have your InfluxDB credentials configured in the CLI, you can install this template with:

influx apply -u https://raw.githubusercontent.com/influxdata/community-templates/master/algorithmia/algorithmia.yml

Included Resources

This template includes the following:

  • 1 Bucket: insights
  • 1 Dashboard: Algorithmia - ML Model Performance Metrics
  • 1 Telegraf configuration

Setup Instructions

This template allows you to monitor and visualize metrics from Algorithmia when a model that has Insights enabled is queried. Each JSON payload from Algorithmia includes operational metrics and inference-related metrics that can be analyzed in InfluxDB.

  1. From your Algorithmia instance, configure Algorithmia Insights to connect to a Kafka broker and specify a Kafka topic.
  2. Start Telegraf using the configuration included in this template.
  3. In Algorithmia, instrument your algorithm and publish a version that has Insights enabled.
  4. When your model is queried in Algorithmia, it will send metrics to InfluxDB where you can monitor and visualize relevant model performance metrics.

The Telegraf configuration requires the following environment variables to be defined:

  • INFLUX_HOST - The URL for your InfluxDB host.

  • INFLUX_TOKEN - The token with the permissions to read Telegraf configs and write data to the telegraf bucket. You can use your operator token to get started.

  • INFLUX_ORG - The name of your organization in InfluxDB.

  • KAFKA_BROKER - The URL for your Kafka broker.

  • KAFKA_TOPIC - The Kafka topic that Algorithmia Insights is configured to use.

You MUST set these environment variables before running Telegraf using something similar to the following commands:

  • This is the URL for your InfluxDB host and should be in the following format:

    • export INFLUX_HOST=http://1.2.3.4:8086/

  • This can be found on the Load Data > Tokens page in InfluxDB:

    • export INFLUX_TOKEN=TOKEN

  • Your organization name can be found on the Settings page in InfluxDB:

    • export INFLUX_ORG=my_org

  • This is the URL for your Kafka broker and should be in the following format:

    • export KAFKA_BROKER=1.2.3.4:9092

  • The Kafka topic should be the same topic that you configured in Algorithmia:

    • export KAFKA_TOPIC=insights

Customizations

Note that this dashboard uses the duration_milliseconds metric, which is one of the operational metrics that is included in the default JSON payload from Algorithmia. The other metrics used in the dashboard (risk_score and approved) were defined by the algorithm developer and will be different depending on your particular algorithm and use case.

Additional metrics can include any user-defined, inference-related metrics that are specified in Algorithmia by the algorithm developer then analyzed in InfluxDB.

Contact

Author: Kristopher Overholt, https://algorithmia.com/

Github: @koverholt