THESE DOCS ARE DEPRECATED SEE ActionML.com/docs
The Guides are moved
The markdown templates are now in https://github.com/actionml/docs.actionml.com. Changes there are automatically published to the live site: actionml.com/docs. Please make any PRs to that new repos.
ActionML maintains a fork of PredictionIO beginning with v0.9.6. We also invented and maintain several templates including the Universal Recommender. This is the github version of our documentation also found on ActionML.com.
- Setup, installing services on 1 standalone machine Not recommended except for very small datasets on a development machine or as a first step towards a distributed Spark setup.
- Setup, installing services on 3 clustered machines Good for a dev/experimental setup with small-ish datasets.
- Setup for a fully distributed cluster The ultimate in scalability with all services split out in their own clusters for independent scaling and maintenance. For very high load and large datasets or for multi-tenancy.
- Scaling beyond the 3-machine setup Issues around horizontal scaling of PIO.
- Distributed Services Setup The minimum needed to split EventServers from PredictionServers and run with completely separate component services like Spark and Elasticsearch.
- Installing ActionML's PredictionIO v0.9.6 This is required for ActionML templates including the Universal Recommender
- Starting and stopping a cluster What you need to do to manually start and stop a fully distributed cluster
- PIO Command Line Interface Cheatsheet