Apache PredictionIO

1
What Is PredictionIO ?

• A Machine Learning Server
• Uses Customizable Model Templates
• Powered By Apache Spark
• Offers SDK's for Java, PHP, Python, Ruby
• Apache 2 License Open Source
• Query in Real Time

2
PredictionIO Architecture
3
PredictionIO Architecture

• PredictionIO Contains
• PredictionIO Platform
• Stack For Engine Management
• Event Server
• Process Events / Data From Multiple Sources
• Template Gallery
• Source Engine Templates

4
PredictionIO Event Server

• Event Server Collects App Data
• Sends Data to Model Engines
• Real Time or Batch
• Supports Multiple Apps
• Send Data via HTTP to Server
• Send Data via SDK's to Server

5
PredictionIO Engine

• Engines Make Predictions
• Deployed As Web Service
• Responds Via REST API
• Uses Training Data to Create Models
• See Template Gallery For Engines

6
PredictionIO Dependencies

• Scala
• Apache Spark 2.0.2
• Hadoop 2.6.5 ( Optional )
• Java 8
• One Of The Following
• PostgreSQL 9.1
• MySQL 5.1
• Apache HBASE 0.98.5 / ElasticSearch 1.7.6

7
PredictionIO Configuration

• PredictionIO Has Three Repositories
• Event, Meta, Model Data
• Storage Can Be
• PostreSQL ( All Data )
• HBase ( Event Data )
• ElasticSearch ( Meta Data )
• MySQL ( All Data )
• Local File System ( Model Data )
• HDFS ( Model Data )
• AWS S3 ( Model Data )

8
PredictionIO Engine

• Download Engine Template
• Using poi Tool
• Build Updates Engine
• Train Creates Predictive Model
• Deploy As Web Service
• Use Existing Template (or)
• Customize Template (or)
• Build New Template

9
PredictionIO Engine Development

• Engines Use D-A-S-E components
• Data Source / Data Preparator
• Algorithm ( ML )
• Serving ( Prediction Queries )
• Evaluation Metrics ( Accuracy ? )
• Deploy As Web Service
• Serves Apps Via REST API

10
PredictionIO Engine Development
11
Available Books

• See Big Data Made Easy
• Apress Jan 2015
• See Mastering Apache Spark
• Packt Oct 2015
• See Complete Guide to Open Source Big Data
  Stack
• Apress Jan 2018
• Find the author on Amazon
• www.amazon.com/Michael-Frampton/e/B00NIQDOOM/
• Connect on LinkedIn
• www.linkedin.com/in/mike-frampton-38563020

12
Connect

• Feel free to connect on LinkedIn
• www.linkedin.com/in/mike-frampton-38563020
• See my open source blog at
• open-source-systems.blogspot.com/
• I am always interested in
• New technology
• Opportunities
• Technology based issues
• Big data integration