Learning Salesforce Einstein

Mohith Shrivastava

• Publisher: Packt Publishing
• Year: 2017
• eBook ISBN: 9781787125285
• Edition: 1
• Topic: Ciencia de la computación
• Subtopic: Inteligencia empresarial

Incorporate the power of Einstein in your Salesforce applicationAbout This Book• Make better predictions of your business processes using prediction and predictive modeling• Build your own custom models by leveraging PredictionIO on the Heroku platform• Integrate Einstein into various cloud services to predict sales, marketing leads, insights into news feeds, and moreWho This Book Is ForThis book is for developers, data scientists, and Salesforce-experienced consultants who want to explore Salesforce Einstein and its current offerings. It assumes some prior experience with the Salesforce platform.What You Will Learn• Get introduced to AI and its role in CRM and cloud applications• Understand how Einstein works for the sales, service, marketing, community, and commerce clouds• Gain a deep understanding of how to use Einstein for the analytics cloud• Build predictive apps on Heroku using PredictionIO, and work with Einstein Predictive Vision Services• Incorporate Einstein in the IoT cloud• Test the accuracy of Einstein through Salesforce reporting and Wave analyticsIn DetailDreamforce 16 brought forth the latest addition to the Salesforce platform: an AI tool named Einstein. Einstein promises to provide users of all Salesforce applications with a powerful platform to help them gain deep insights into the data they work on.This book will introduce you to Einstein and help you integrate it into your respective business applications based on the Salesforce platform. We start off with an introduction to AI, then move on to look at how AI can make your CRM and apps smarter. Next, we discuss various out-of-the-box components added to sales, service, marketing, and community clouds from salesforce to add Artificial Intelligence capabilities. Further on, we teach you how to use Heroku, PredictionIO, and the force.com platform, along with Einstein, to build smarter apps.The core chapters focus on developer content and introduce PredictionIO and Salesforce Einstein Vision Services. We explore Einstein Predictive Vision Services, along with analytics cloud, the Einstein Data Discovery product, and IOT core concepts. Throughout the book, we also focus on how Einstein can be integrated into CRM and various clouds such as sales, services, marketing, and communities.By the end of the book, you will be able to embrace and leverage the power of Einstein, incorporating its functions to gain more knowledge. Salesforce developers will be introduced to the world of AI, while data scientists will gain insights into Salesforce's various cloud offerings and how they can use Einstein's capabilities and enhance applications.Style and approachThis book takes a straightforward approach to explain Salesforce Einstein and all of its potential applications. Filled with examples, the book presents the facts along with seasoned advice and real-world use cases to ensure you have all the resources you need to incorporate the power of Einstein in your work.

Building Smarter Apps Using PredictionIO and Heroku

If you recall in Chapter 1, Introduction to AI, we covered the basics of machine learning and covered a simple experiment by integrating Google's Prediction API with the Force.com platform. The infrastructure and the web service was all maintained by Google, and we just consumed it.

In this chapter, we will cover PredictionIO which is a part of Einstein's offering exclusive for developers to build machine learning algorithms with Salesforce/external data. PredictionIO abstracts the complexity of setting a machine learning system to build predictive applications. It helps us simplify by setting the infrastructure and responds to dynamic queries in real time once deployed as a web service. It is also used internally by Salesforce and various other open source apps. Since it is open source, developers can use this to build smarter apps with the Salesforce data.

This chapter will cover the basics of PredictionIO and how to deploy a machine learning algorithm and engine offered by PredictionIO on the Heroku Cloud. Heroku is a part of the Salesforce App Cloud offering that lets you run, deploy, and manage applications written in Java, Scala, Node.js, Ruby, Go, and PHP without worrying about the infrastructure. The Heroku Cloud makes an excellent developer experience with its command-line interface (CLI) and integration support for git/GitHub (web-based version control repository) and Docker (Docker is an open source project that automates the deployment of applications inside software containers).

The chapter starts with an Introduction to PredictionIO, architecture, and how to run it locally on MAC/LINUX. Once we understand the different components PredictionIO provides, we will explore how we can deploy a machine learning algorithm and engine on the Heroku infrastructure. This chapter is exclusively targeted toward developers and data scientists, and the prerequisite for this chapter is that you have an understanding of the steps involved in machine learning, which was covered in Chapter 1, Introduction to AI. Familiarity with Java/Scala and using git as a version control system, understanding the basic Terminal commands of MAC/LINUX, and also some basics of Hadoop will help accelerate the understanding of concepts. To gain maximum benefit from the content, it is advisable to use the MAC/LINUX machine for installation and try it out hands-on to gain the necessary practical insight.

In this chapter, we will cover the following topics:

• Introduction to PredictionIO
• Architecture and Integration with web/mobile applications
• Installation of PredictionIO
• Getting started with PredictionIO
• PredictionIO DASE components and customization of Engine
• Deploying PredictionIO on Heroku

Introduction to PredictionIO

The process of building machine learning algorithms requires extensive knowledge of mathematics and statistics. Some smart data scientists have put together open source libraries, such as MLlib (https://spark.apache.org/docs/latest/mllib-guide.html), and various tools and engines on the internet to make this process simpler. However, it is still resource intensive to put together all of this for a machine learning project. This is where PredictionIO makes it simpler by abstracting all these dependencies.

As per Apache, "PredictionIO is an open source Machine Learning Server built on top of state-of-the-art open source stack, that enables developers to manage and deploy production-ready predictive services for various kinds of machine learning tasks."

It provides full stack machine learning systems to develop and deploy machine learning capabilities.

PredictionIO platform components

The PredictionIO platform components consist of the following components:

• PredictionIO Framework: This provides a stack to build and deploy engines with machine learning algorithms. PredictionIO uses Apache Spark (http://spark.apache.org/) for data processing and MLlib (https://spark.apache.org/docs/latest/mllib-guide.html) to assist with predictive modeling. In this chapter, we will look into the installation of the engine in detail.

Apache Spark is a fast, in-memory data processing engine with development APIs to allow data workers to execute streaming, machine learning, or SQL. It is known to run 100 times faster than Hadoop MapReduce.

• Event Server: This is a machine learning analytics layer used by the PredictionIO platform to collect events from multiple systems. This layer can use Apache HBase (Apache HBase is a data store that runs on top of the Hadoop Distributed File System (HDFS); Hadoop is a framework to handle large datasets in a distributed computing environment) or the Java Database Connectivity (JDBC) backend as its data store. We will look into details on how to set an Event Server later. An Event Server will be a REST endpoint that listens for various events.
• Template Gallery and software development kit (SDKs): Predefined Templates Gallery (https://predictionio.incubator.apache.org/gallery/template-gallery/) for developers to provide a q...

Table of contents

• Title Page
• Copyright
• Credits
• About the Author
• About the Reviewer
• www.PacktPub.com
• Customer Feedback
• Preface
• Introduction to AI
• Role of AI in CRM and Cloud Applications
• Building Smarter Apps Using PredictionIO and Heroku
• Product Recommendation Application using PredicitionIO and Salesforce App Cloud
• Salesforce Einstein Vision
• Building Applications Using Einstein Vision and Salesforce Force.com Platform
• Einstein for Analytics Cloud
• Einstein and Salesforce IoT Cloud Platform
• Measuring and Testing the Accuracy of Einstein