Book Image

Practical Data Science Cookbook

By : Tony Ojeda, Sean Patrick Murphy, Benjamin Bengfort, Abhijit Dasgupta
Book Image

Practical Data Science Cookbook

By: Tony Ojeda, Sean Patrick Murphy, Benjamin Bengfort, Abhijit Dasgupta

Overview of this book

<p>As increasing amounts of data is generated each year, the need to analyze and operationalize it is more important than ever. Companies that know what to do with their data will have a competitive advantage over companies that don't, and this will drive a higher demand for knowledgeable and competent data professionals.</p> <p>Starting with the basics, this book will cover how to set up your numerical programming environment, introduce you to the data science pipeline (an iterative process by which data science projects are completed), and guide you through several data projects in a step-by-step format. By sequentially working through the steps in each chapter, you will quickly familiarize yourself with the process and learn how to apply it to a variety of situations with examples in the two most popular programming languages for data analysis—R and Python.</p>
Table of Contents (18 chapters)
Practical Data Science Cookbook
Credits
About the Authors
About the Reviewers
www.PacktPub.com
Preface
Index

Understanding the optimization process


In this abstract recipe, we will talk about the steps required to optimize software.

How to do it…

Perform the following steps to learn about the optimization process that we will be employing:

  1. Establish the baseline performance of your existing code in terms of the relevant parameters (execution time, memory consumed, peak I/O bandwidth, and so on).

  2. Determine the performance targets or system constraints. One of the seven habits of highly effective people was to always start with the end in mind, and optimization is no different. How quickly must the code get executed? What is the longest acceptable time to complete the processes? What is the maximum amount of memory that can be consumed by your software? How accurate must the computed results be? Don't forget about the fidelity of the results.

  3. Set up a development and measurement environment that allows you to easily and rapidly measure and record the relevant performance characteristics. The easier it...