Book Image

The Applied Data Science Workshop - Second Edition

By : Alex Galea
Book Image

The Applied Data Science Workshop - Second Edition

By: Alex Galea

Overview of this book

From banking and manufacturing through to education and entertainment, using data science for business has revolutionized almost every sector in the modern world. It has an important role to play in everything from app development to network security. Taking an interactive approach to learning the fundamentals, this book is ideal for beginners. You’ll learn all the best practices and techniques for applying data science in the context of real-world scenarios and examples. Starting with an introduction to data science and machine learning, you’ll start by getting to grips with Jupyter functionality and features. You’ll use Python libraries like sci-kit learn, pandas, Matplotlib, and Seaborn to perform data analysis and data preprocessing on real-world datasets from within your own Jupyter environment. Progressing through the chapters, you’ll train classification models using sci-kit learn, and assess model performance using advanced validation techniques. Towards the end, you’ll use Jupyter Notebooks to document your research, build stakeholder reports, and even analyze web performance data. By the end of The Applied Data Science Workshop, you’ll be prepared to progress from being a beginner to taking your skills to the next level by confidently applying data science techniques and tools to real-world projects.
Table of Contents (8 chapters)

Machine Learning Process

Machine learning (ML) lies at the heart of data science. It is an umbrella term for a huge set of algorithms that find and model patterns in data. These algorithms can be broken down into various categories, such as supervised, unsupervised, and reinforcement learning.

In supervised problems, we have access to a historical view of labeled records and fit models to predict them—for example, blood test data that's been labeled with the test result. In unsupervised problems, there is no such data available, and labels may need to be created using clustering techniques. In later sections, we will break these down in more detail and work with examples of each.

Reinforcement learning is concerned with maximizing a reward function through an iterative process, such as a simulation. Similar to the other types of learning algorithms, there's a wide range of problems that reinforcement learning can be applied to, such as teaching a robot how to...