Neural Data Science

A Primer with MATLAB® and Python™

Erik Lee Nylen, Pascal Wallisch

• Publisher: Academic Press
• Year: 2017
• Print ISBN: 9780128040430
• eBook ISBN: 9780128040980
• Topic: Biological Sciences
• Subtopic: Science Research & Methodology
• Index: Biological Sciences

A Primer with MATLAB® and Python™ present important information on the emergence of the use of Python, a more general purpose option to MATLAB, the preferred computation language for scientific computing and analysis in neuroscience.This book addresses the snake in the room by providing a beginner's introduction to the principles of computation and data analysis in neuroscience, using both Python and MATLAB, giving readers the ability to transcend platform tribalism and enable coding versatility.- Includes discussions of both MATLAB and Python in parallel- Introduces the canonical data analysis cascade, standardizing the data analysis flow- Presents tactics that strategically, tactically, and algorithmically help improve the organization of code

Part I

Foundations

Outline

Chapter 1 Philosophy

Chapter 2 From 0 to 0.01

Chapter 1

Philosophy

Abstract

In this chapter, we outline the current state of neuroscience and the need to integrate the emerging field of data science with neuroscience, yielding the field of neural data science. We detail how this new field contrasts with classic computational neuroscience and also give a brief history of data science, showing why it needs refinements to calibrate it to the special needs arising when handling neural data. We describe the role of neural data scientists in the neuroscience of the future and compare it to that of full-stack engineers. Finally, we justify covering both MATLAB and Python jointly in this book and explain why we did not deem it necessary to cover other languages.

Keywords

Neuroscience; data science; neural data science; data; Plato’s cave; information; Python; MATLAB; coding; computational neuroscience

Before we get started—and perhaps even bogged down—with specific algorithms and how to implement them in code, we want to lay out how we conceptualize the emerging field of neural data science (NDS) as well as the relevant challenges within neuroscience that NDS can help address. This is important so as to not get lost in the random forest.

What Is Data Science?

The field of data science emerged in the early 21st century as a distinct scientific discipline. It can be thought of as a fusion of concepts from computer science, statistics, and machine learning in conjunction with the availability of both an abundance of data as well as sufficient computational power that allows for the implementation of these concepts on this data, in code. (Hilbert & López, 2011). Data science has plenty of historical antecedents that can be traced well into the 20th century. However, people that were doing “data science” then would be surprised to learn that they were doing so at the time—instead, they saw themselves as doing statistics (Tukey, 1977), working on problems of communication theory (Shannon, 1948), computational theories (Turing, 1938), or cryptanalysis (Good, 1979), to name just a few.

The term “data scientist” (a practitioner of data science) was coined by Jeff Hammerbacher and D.J. Patil in 2008 (Watson, 2013). It has been pointed out that the term data science is somewhat unfortunate, as there is no science without data, in this sense every science is a data science (Wallisch, 2013, 2014). As will hopefully become clear shortly, “neural metralogy” would be a more suitable term. However, data science in the more narrow sense describes the relentless pursuit of understanding and explaining the structure of data, of making predictions from this understanding using any means necessary, mostly by using algorithms and principles from computer science, statistics, and machine learning. Data science explicitly transcends classical discipline boundaries, e.g., a biologist is usually only interested in data from biology, a chemist only interested in data relevant to questions in chemistry, and so on. In data science, the data themselves take center stage, within an explicitly data-centric framework. In its strong form, the source of the data does not matter, as data are data. The first time that a data scientist has been featured in popular fiction was, to our knowledge, in season 4 of the Netflix television show “House of Cards,” though individuals such as Nate Silver have prominently—if not always successfully—provided insights into data from tribalist entertainment media, in particular, sports and politics. As of 2016, data science is still rapidly growing and changing, but a consensus about its core concepts has now been established.

What Is Neural Data Science?

If the statements in our previous section are true (we believe they are), neural data science appears to be a contradiction in terms. How can one have a general data science that is deliberately indifferent about the source of its data but is at the same time bounded by neural data? This ...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright
• Dedication
• Biography
• Preface
• How to Use This Book
• Part I: Foundations
• Part II: Neural Data Analysis
• Part III: Going Beyond the Data
• Appendix A. MATLAB to Python (Table of Equivalences)
• Appendix B. Frequently Made Mistakes
• Appendix C. Practical Considerations, Technical Issues, Tips and Tricks
• Glossary (Including Additional Python and MATLAB Packages and Examples)
• Bibliography
• Index