Book Image

Practical Machine Learning Cookbook

By : Atul Tripathi
Book Image

Practical Machine Learning Cookbook

By: Atul Tripathi

Overview of this book

Machine learning has become the new black. The challenge in today’s world is the explosion of data from existing legacy data and incoming new structured and unstructured data. The complexity of discovering, understanding, performing analysis, and predicting outcomes on the data using machine learning algorithms is a challenge. This cookbook will help solve everyday challenges you face as a data scientist. The application of various data science techniques and on multiple data sets based on real-world challenges you face will help you appreciate a variety of techniques used in various situations. The first half of the book provides recipes on fairly complex machine-learning systems, where you’ll learn to explore new areas of applications of machine learning and improve its efficiency. That includes recipes on classifications, neural networks, unsupervised and supervised learning, deep learning, reinforcement learning, and more. The second half of the book focuses on three different machine learning case studies, all based on real-world data, and offers solutions and solves specific machine-learning issues in each one.
Table of Contents (21 chapters)
Practical Machine Learning Cookbook
Credits
About the Author
About the Reviewer
www.PacktPub.com
Customer Feedback
Preface
14
Case Study - Forecast of Electricity Consumption

Recurrent neural networks - predicting periodic signals


Oscillators are circuits that produce specific, periodic waveforms such as square, triangular, sawtooth, and sinusoidal. In order to generate output, oscillators generally use some form of active device-lamp, which is surrounded by resistors, capacitors, and inductors. Two main classes of oscillators are relaxation and sinusoidal. Triangular, sawtooth and other non-sinusoidal waveforms are generated using relaxation oscillators, while sinusoidal oscillators consist of amplifiers with external components to generate oscillation. Normally, no harmonics are present in pure sine waves and they consist of a single frequency.

Getting ready...

The task is to predict a cosine from a noisy sine wave. 5Hz frequency waves are used for the sine wave with some normally distributed noise and a smooth cosine wave. The dataset created is a set of 10 sequences, each of which consists of 40 observations.

How to do it...

The following packages need to be...