Название: A First Course in Statistical Learning: With Data Examples and Python Code
Автор: Johannes Lederer
Издательство: Springer
Год: 2025
Страниц: 293
Язык: английский
Формат: pdf (true)
Размер: 17.9 MB

This textbook introduces the fundamental concepts and methods of statistical learning. It uses Python and provides a unique approach by blending theory, data examples, software code, and exercises from beginning to end for a profound yet practical introduction to statistical learning. The book consists of three parts: The first one presents data in the framework of probability theory, exploratory data analysis, and unsupervised learning. The second part on inferential data analysis covers linear and logistic regression and regularization. The last part studies Machine Learning with a focus on support-vector machines and Deep Learning. Each chapter is based on a dataset, which can be downloaded from the book's homepage.

In addition, the book has the following features:
A careful selection of topics ensures rapid progress.
An opening question at the beginning of each chapter leads the reader through the topic.
Expositions are rigorous yet based on elementary mathematics.
More than two hundred exercises help digest the material.
A crisp discussion section at the end of each chapter summarizes the key concepts and highlights practical implications.

Data Science is the art of turning data into information. Conducting data science on the basis of statistical principles is called statistical learning. Statistical learning establishes systematic rules for data and describes deviations from these rules. This book introduces the foundations of statistical learning and the most common methods. It consists of three parts:

Data are the basis for statistical learning. Our mathematical framework for data is probability theory, which models variations in the data in line with our intuitive notion of chance. These data are then described by basic summary statistics (averages, number of samples, ...), visualizations (histograms, q-q plots, ...), and unsupervised learning (PCA, k-means, ...).
• Chapter 1: Fundamentals of Data
• Chapter 2: Exploratory Data Analysis
• Chapter 3: Unsupervised Learning
Inferential Data Analyses draw conclusions about the data-generating process. The hearts of these analyses are statistical models, such as linear- and logistic regression models.
• Chapter 4: Linear Regression
• Chapter 5: Logistic Regression
• Chapter 6: Regularization
Machine Learning makes predictions about new data. These predictions can involve statistical models, but the models are not the main focus.
• Chapter 7: Support-Vector Machines
• Chapter 8: Deep Learning

Why Python?
A number of different programming languages are used in data science, including Python, Julia, Java, R, and MATLAB. Each of these languages has its distinct advantages and disadvantages, and to become an expert data scientist, you may need to master more than one of them. But the programming language that is most suitable for beginners is Python: it is open-source, flexible yet comparably easy to learn, and comes with a large number of software libraries and tutorials. These features have made Python by far the most popular programming language in data science—as well as one of the most popular programming languages overall.

The code snippets introduce the Python functionality needed for our data analyses. Additionally, some general Python tricks are introduced where appropriate. If you require further details on specific commands, functions, or modules, refer to the many online documentations. More generally, the book is not designed as a self-contained introduction to Python; if you are entirely new to Python, consider taking one of the many excellent online tutorials first.

What Python Modules are Required?
Python and nine well-established modules are sufficient: Python 3.10.6, numpy 1.23.5, pandas 1.5.2, matplotlib 3.6.2, scipy 1.9.3, seaborn 0.12.0, sklearn 1.1.2, plotly 5.10.0, torch 1.13.0, torchvision 0.14.0.

For Whom is This Book?
This textbook is for everyone who wants to understand and apply the fundamental concepts and methods of statistical learning. Typical readers are third- and fourth-year undergraduate and first- and second-year graduate students in data-intensive fields such as computer science, biology, psychology, business, and engineering, or they are graduates preparing for their job interviews.