Book Image

Mastering Predictive Analytics with R - Second Edition

By : James D. Miller, Rui Miguel Forte
Book Image

Mastering Predictive Analytics with R - Second Edition

By: James D. Miller, Rui Miguel Forte

Overview of this book

R offers a free and open source environment that is perfect for both learning and deploying predictive modeling solutions. With its constantly growing community and plethora of packages, R offers the functionality to deal with a truly vast array of problems. The book begins with a dedicated chapter on the language of models and the predictive modeling process. You will understand the learning curve and the process of tidying data. Each subsequent chapter tackles a particular type of model, such as neural networks, and focuses on the three important questions of how the model works, how to use R to train it, and how to measure and assess its performance using real-world datasets. How do you train models that can handle really large datasets? This book will also show you just that. Finally, you will tackle the really important topic of deep learning by implementing applications on word embedding and recurrent neural networks. By the end of this book, you will have explored and tested the most popular modeling techniques in use on real- world datasets and mastered a diverse range of techniques in predictive analytics using R.
Table of Contents (22 chapters)
Mastering Predictive Analytics with R Second Edition
Credits
About the Authors
About the Reviewer
www.PacktPub.com
Customer Feedback
Preface
8
Dimensionality Reduction
Index

Stochastic gradient descent


In the models we've seen so far, such as linear regression, we've talked about a criterion or objective function that the model must minimize while it is being trained. This criterion is also sometimes known as the cost function. For example, the least squares cost function for a model can be expressed as:

We've added a constant term of ½ in front of this for reasons that will become apparent shortly. We know from basic differentiation that when we are minimizing a function, multiplying the function by a constant factor does not alter the value of the minimum value of the function. In linear regression, just as with our perceptron model, our model's predicted are just the sum of a linear weighted combination of the input features. If we assume that our data is fixed and that the weights are variable and must be chosen so as to minimize our criterion, we can treat the cost function as being a function of the weights:

We have used the letter w to represent the model...