PredictionIO Ruby SDK Simplified

The Ruby SDK provides a simple wrapper for the PredictionIO API. It allows you to quickly record your users' behavior and retrieve personalized predictions for them.

Why use this sdk over the official one?

The official SDK forces an async architecture on you. It spins threads regardless of whether you use the blocking or async methods. We believe that async requests should be outside the scope of the client and be handled by the consumer. Also, we found that it doesn't handle transient failure very well.
The official SDK uses raw Net::HTTP. This client uses Faraday which you may configure any way you like (see examples below).

Documentation

Please see the PredictionIO App Integration Overview to understand how the SDK can be used to integrate PredictionIO Event Server and Engine with your application.

Installation

Ruby 1.9.3+ required!

The module is published to RubyGems and can be installed directly by:

gem install predictionio-simple

Or using Bundler with:

gem 'predictionio-simple', '~> 0.10.0.1'

Sending Events to Event Server

Please refer to Event Server documentation for event format and how the data can be collected from your app.

Instantiate Event Client and connect to PredictionIO Event Server

require 'predictionio'

# Define environment variables.
ENV['PIO_EVENT_SERVER_URL'] = 'http://localhost:7070'
ENV['PIO_ACCESS_KEY'] = 'YOUR_ACCESS_KEY' # Find your access key with: `$ pio app list`.

# Create PredictionIO event client.
client = PredictionIO::EventClient.new(ENV['PIO_ACCESS_KEY'], ENV['PIO_EVENT_SERVER_URL'])

# Or optionally pass a block to configure faraday
client = PredictionIO::EventClient.new(ENV['PIO_ACCESS_KEY'], ENV['PIO_EVENT_SERVER_URL']) do |faraday|
  faraday.response :logger                  # log requests to STDOUT
end

Create a `$set` user event and send it to Event Server

client.create_event(
  '$set',
  'user',
  user_id
)

Create a `$set` item event and send it to Event Server

client.create_event(
  '$set',
  'item',
  item_id,
  { 'properties' => { 'categories' => ['Category 1', 'Category 2'] } }
)

Create a user 'rate' item event and send it to Event Server

client.create_event(
  'rate',
  'user',
  user_id, {
    'targetEntityType' => 'item',
    'targetEntityId' => item_id,
    'properties' => { 'rating' => 10 }
  }
)

Query PredictionIO Engine

Connect to the Engine:

# Define environmental variables.
ENV['PIO_ENGINE_URL'] = 'http://localhost:8000'

# Create PredictionIO engine client.
client = PredictionIO::EngineClient.new(ENV['PIO_ENGINE_URL'])

# Or optionally pass a block to configure faraday
client = PredictionIO::EngineClient.new(ENV['PIO_ENGINE_URL']) do |faraday|
  faraday.response :logger                  # log requests to STDOUT
end

Send a prediction query to the engine and get the predicted result:

# Get 5 recommendations for items similar to 10, 20, 30.
response = client.send_query(items: [10, 20, 30], num: 5)

Issue Tracker

Use GitHub Issues.

Contributing

We follow the [git-flow] (http://nvie.com/posts/a-successful-git-branching-model/) model where all active development goes to the develop branch, and releases go to the master branch. Pull requests should be made against the develop branch and include relevant tests, if applicable.

License

Apache License 2.0.

predictionio-simple

Runtime