Hypertune
  • Introduction
  • Getting Started
    • Set up Hypertune
    • Next.js (App Router) quickstart
    • Next.js (Pages Router) quickstart
    • React quickstart
    • Remix quickstart
    • Gatsby quickstart
    • Vue quickstart
    • Nuxt quickstart
    • Node.js quickstart
    • React Native quickstart
    • JavaScript quickstart
    • Python quickstart
    • Rust quickstart
    • Go quickstart
    • Web quickstart
    • GraphQL quickstart
  • Example apps
    • Next.js and Vercel example app
  • Concepts
    • Architecture
    • Project
    • Schema
    • Flag lifecycle
    • Logic
    • Variables
    • Splits
    • A/B tests
    • Staged rollouts
    • Multivariate tests
    • Machine learning loops
    • Events
    • Funnels
    • Hypertune Edge
    • Reduction
    • SDKs
    • GraphQL API
    • Git-style version control
    • App configuration
  • Use Cases
    • Feature flags and A/B testing
    • Landing page optimization
    • In-app content management
    • Pricing plan management
    • Permissions, rules and limits
    • Optimizing magic numbers
    • Backend configuration
    • Product analytics
  • Integrations
    • Vercel Edge Config integration
    • Google Analytics integration
    • Segment integration
    • Webhooks
      • Creating webhooks
      • Handling webhooks
  • SDK Reference
    • Installation
    • Type-safe client generation
    • Initialization
    • Build-time logic snapshot
    • Hard-coded fallbacks
    • Local-only, offline mode
    • Hydrate from your own server
    • Wait for server initialization
    • Provide targeting attributes
    • Local, synchronous evaluation
    • Remote logging
    • Getting flag updates
    • Serverless environments
    • Vercel Edge Config
    • Custom logging
    • Shutting down
Powered by GitBook
On this page
  1. Concepts

Machine learning loops

PreviousMultivariate testsNextEvents

Last updated 10 months ago

You can convert any Test into an ML loop split.

A machine learning loop requires a goal function, expressed as a formula of your .

And instead of setting fixed traffic allocation percentages for each arm, Hypertune automatically and continuously learns the best arm for each dimension. If the split has a payload event type, the payload event fields are used as "features" for the machine learning model.

For example, you can set up a machine learning loop to personalize the headline on your landing page to each unique visitor, to maximize sign ups.

split
event types