Data Science from Scratch / 2nd Edition: First Principles with Python by Joel Grus

Joel Grus makes a deliberately awkward choice: no scikit-learn, no pandas shortcuts. In Data Science from Scratch you implement linear regression, gradient descent, k-means and neural networks in plain Python, because the fastest route to understanding a library is to write the thing it replaces.

What the book covers

• A Python crash course, then linear algebra, statistics and probability implemented from scratch
• Hypothesis testing, gradient descent, and getting and cleaning data
• Working with data: exploration, rescaling, dimensionality reduction
• Machine learning fundamentals — k-nearest neighbours, naive Bayes, regression, decision trees
• Neural networks and deep learning, built up layer by layer
• Clustering, natural language processing, network analysis and recommender systems
• MapReduce, databases, and where to go next with real libraries

Who it’s for

Programmers moving into data science, students who want the mechanics rather than the API surface, and anyone who has used a model and never quite known what it did. The second edition is fully updated to Python 3 with type annotations throughout.

Edition details

• Author: Joel Grus
• Publisher: O’Reilly Media
• Edition: 2nd Edition
• Format: Paperback
• ISBN: 9781492041139
• Language: English

Ships free within the United States from Shelfcove.