This book by Ashish Kumar, a data scientist at Tiger Analytics (India), is a comprehensive book on Predictive Analytics and Python for aspiring data scientists. A commendable content review by Matt Hollingsworth (Co-founder Global Dressage Analytics, Netherlands) and a promising foreword by Pradeep Gulipalli (Co-founder, Tiger Analytics, India), makes it a must read for any Python aspirants.
Primarily, this an intermediate level book, catering to professionals or advanced engineering students who know the basics of python and advanced mathematics. It is a well-organized, up-to-date as well as relevant for today’s times. There are nine chapters in the book. The first few chapters explain data exploration and cleaning while the subsequent chapters cover predictive modelling algorithms. Broadly, each chapter covers the math behind the model, different types of the model, implementation and interpretation of the results.
Dissecting each chapter, we found that the first chapter starts off by giving practical use-cases of predictive modelling, cultivating and capturing the initial interest of the reader in the subject and providing the bait to continue reading.
Chapter two onwards, takes a dive into the ‘hands-on’ stuff as it deals with organizing the data with Python and preparing a good base for the upcoming predictive analytics part. This chapter is highly useful for beginners as it gives them a good start, but portions of this chapter can be skipped by advanced readers who already have some knowledge of Python.
Now, chapter three is about “data wrangling” – a fancy term for extracting relevant part of the data in an efficient manner. The beauty of this chapter lies in the transitions between the mathematics and the python coding. The transitions are so smooth that even inexperienced readers from either math or programming backgrounds can grasp the concepts. By the end, this chapter starts moving towards implementations.
In chapter four, the author covers very little of python but sets up a strong base for maths as he explains the math behind the concepts. Hence, users well versed with math may skip portions of this chapter.
However, the actual modelling starts from Chapter five onwards, which deals with linear regression while Chapter six is on logistic regression. Yet again, all the algorithms are introduced in math first followed by python coding. Moreover, the readers are encouraged to do the modelling and write python code for implementing the models. Thus, by the end of chapter six, the reader has a very good hold on python as well as its underlying mathematics.
While Chapters seven & eight look at two categories of predictive modeling through unsupervised and supervised classification algorithm respectively; chapter nine covers the best practices for implementation – right from writing clean code to statistical tests.
Overall, this is a very ‘hands-on’ or self-help book that gives tasks and exercises. Moreover, what works well for this book is the fact that, the author takes us through commonly encountered business scenarios and quotes examples from publicly available use cases such as LinkedIn with which we can connect easily; thus, ensuring a steady flow of interest for reading.
Additionally, the statistics and math behind the models are clearly explained. Also, several times, the author has compared two different models and explained their individual relevance and benefits. Besides, the appendix of this book divulges the readers to several Python libraries while mentoring them for best practices in Python.
Title: Learning Predictive Analytics with Python
Author: Ashish Kumar
Publisher: Packt Publishing
Formats: Ebook / Print
1st Edition: February 2016
Ebook ISBN: 978-1-78398-327-8 | ISBN 10: 1-78398-327-2
Print ISBN: 9781847199744
You can buy it online from:
Try deep learning using MATLAB