Packt+ | Advance your knowledge in tech

You're reading from Apache Spark 2.x Machine Learning Cookbook Over 100 recipes to simplify machine learning model implementations with Spark

Product type Paperback

Published in Sep 2017

Publisher Packt

ISBN-13 9781783551606

Length 666 pages

Edition 1st Edition

Languages

Scala

Tools

Apache Spark

Concepts

Machine Learning

Authors (5):

Broderick Hall

Meenakshi Rajendran

Shuen Mei

Mohammed Guller

Siamak Amirghodsi

+1 more

View More author details

Table of Contents (14) Chapters

Preface

1. Practical Machine Learning with Spark Using Scala FREE CHAPTER

2. Just Enough Linear Algebra for Machine Learning with Spark

3. Spark's Three Data Musketeers for Machine Learning - Perfect Together

4. Common Recipes for Implementing a Robust Machine Learning System

5. Practical Machine Learning with Regression and Classification in Spark 2.0 - Part I

6. Practical Machine Learning with Regression and Classification in Spark 2.0 - Part II

7. Recommendation Engine that Scales with Spark

8. Unsupervised Clustering with Apache Spark 2.0

9. Optimization - Going Down the Hill with Gradient Descent

10. Building Machine Learning Systems with Decision Tree and Ensemble Models

11. Curse of High-Dimensionality in Big Data

12. Implementing Text Analytics with Spark 2.0 ML Library

13. Spark Streaming and Machine Learning Library

Lasso regression with SGD optimization in Spark 2.0

In this recipe, we will use the housing dataset from the recipes to demonstrate shrinkage with Spark's RDD-based lasso regression LassoWithSGD(), which can select a subset of parameters by setting the other weights to zero (hence eliminating some parameters based on the threshold) while reducing the effect of others (regularization). We emphasize again that ridge regression reduces the parameter weight, but never sets it to zero.

LassoWithSGD(), which is Spark's RDD-based lasso (Least Absolute Shrinkage and Selection Operator) API, a regression method that performs both variable and regularization at the same time in order to eliminate non-contributing explanatory variables (that is, features), therefore enhancing the prediction's accuracy. Lasso, which is based on Ordinary Least Squares (OLS), can be easily to other methods, such as General Liner Methods (GLM).