Joel Grus takes a rare approach: instead of relying on pre-built libraries like pandas or scikit-learn, he teaches you to build everything from scratch—vectors, statistics, probability, gradient descent, and ML models.

1. Build Data Science from First Principles

CS or stats students seeking a deeper foundation Self-taught learners who don’t want shortcuts Developers transitioning into data science Anyone who wants to think like a true data scientist

8. Who Should Read This?

Data Science from Scratch doesn’t just teach you how to do data science. It teaches you why it works, what it’s built on, and how to think critically at every step. > Learn it. Code it. Understand it. Build it from scratch.

Final Insight

Above all, Grus shows that data science is about mental discipline, not tools. Building things yourself forces you to understand deeply and question assumptions.

7. Data Science Is a Way of Thinking

One key chapter is about gradient descent. You’ll learn to define loss functions, compute derivatives, and update parameters step by step — just like training a model manually.

4.Gradient Descent: Learning by Optimization

Python here is not just a programming language — it’s a vehicle for understanding math and logic. You'll code everything manually: dot products, means, correlations, regressions, and more.

3. Python as a Thinking Tool

This book reads like a hands-on course. You’ll cover: <ul><li>Descriptive & inferential statistics</li></ul> <ul><li>Linear algebra & vector math</li></ul> <ul><li>Probability theory</li></ul> <ul><li>Gradients & optimization</li></ul> <ul><li>Supervised learning models</li></ul> <ul><li>Basic recommender systems</li></ul> <ul><li>Graph analysis</li></ul>

2. A Complete Curriculum in Disguise

Rather than abstracting everything away, this book has you build core algorithms from the ground up: <ul><li>Linear regression</li></ul> <ul><li>Logistic regression</li></ul> <ul><li>k-Nearest Neighbors</li></ul> <ul><li>Simple decision trees</li><li> </li><li>Recommendation engines based on similarity</li></ul>

5. Machine Learning Without Magic

You'll explore concepts like: <ul><li>Correlation vs. causation</li></ul> <ul><li>Pitfalls of averages</li></ul> <ul><li>When to use standard deviation vs. interquartile range</li></ul> <ul><li>Hypothesis testing and p-values</li></ul>

6. Statistical Thinking as a Mental Framework

Discover how Data Science from Scratch by Joel Grus teaches you to build every core concept — from statistics to machine learning — using pure Python. No shortcuts, just deep understanding. Ideal for those who want to truly master data science fundamentals.

Data science libraries, frameworks, modules, and toolkits are great for doing data science, but they’re also a good way to dive into the discipline without actually understanding data science. In this book, you’ll learn how many of the most fundamental data science tools and algorithms work by implementing them from scratch.If you have an aptitude for mathematics and some programming skills, author Joel Grus will help you get comfortable with the math and statistics at the core of data science, and with hacking skills you need to get started as a data scientist. Today’s messy glut of data holds answers to questions no one’s even thought to ask. This book provides you with the know-how to dig those answers out.<ul><li>Get a crash course in Python</li><li>Learn the basics of linear algebra, statistics, and probability—and understand how and when they're used in data science</li><li>Collect, explore, clean, munge, and manipulate data</li><li>Dive into the fundamentals of machine learning</li><li>Implement models such as k-nearest Neighbors, Naive Bayes, linear and logistic regression, decision trees, neural networks, and clustering</li><li>Explore recommender systems, natural language processing, network analysis, MapReduce, and databases</li></ul>

6. Statistical Thinking as a Mental Framework

Similar ideas to 6. Statistical Thinking as a Mental Framework

A Mental Model For Productivity