Rust for Data Science: Efficient Data Analysis and Machine Learning

Data is the new gold, but processing it requires not only power but also reliability. The book "Rust for data science" is your guide to a world where the speed of C++ meets the safety of Python, and memory control does not sacrifice performance. Author Alexey Voronov, a practicing data engineer, shows how Rust is becoming a serious alternative to traditional data science tools, especially when it comes to large volumes of data, stream processing, and production deployment.

The book is structured as a practical guide: you will start with the basics of Rust syntax, but by the third chapter you will be writing efficient pipelines for reading, filtering, and aggregating multi-gigabyte CSV files. Special attention is paid to the ecosystem: you will get acquainted with the libraries polars (DataFrame in Rust, faster than Pandas), ndarray for numerical computations, burn and candle for machine learning, as well as rayon and tokio for parallel and asynchronous computing.

• Chapters 1–2. Introduction to Rust: ownership, borrowing, lifetimes — minimum theory, maximum practice for data tasks.
• Chapters 3–4. Working with data: reading/writing CSV, JSON, Parquet; cleaning and transformation using Polars.
• Chapters 5–6. Statistical analysis and visualization (via plotters and export to Plotly).
• Chapters 7–8. Machine learning: linear regression, decision trees, neural networks in Rust.
• Chapters 9–10. Optimization: parallel computing, SIMD, working with GPU via wgpu.
• Chapter 11. Model deployment: WebAssembly for the browser, REST API on Actix, CLI utilities.

This book is intended for:

• Data scientists tired of slow Python scripts and seeking performance without sacrificing code readability.
• Backend developers who want to integrate ML pipelines into high-load systems.
• Data engineers building ETL processes that require reliability and speed.

After reading, you will be able to:

• Write safe, fast, and parallel programs for data processing.
• Create production-ready pipelines that run 10–100 times faster than Python counterparts.
• Integrate Rust code into existing projects via FFI or WebAssembly.
• Use modern machine learning libraries in Rust.

The book assumes basic programming knowledge (Python or C/C++), but does not require prior experience with Rust. All code examples are available in the repository and come with step-by-step instructions. If you want to take your data work to a new level of speed and reliability — this book is for you.