This is the sample project for the book. mlrs is a single, progressively evolving
Cargo workspace that models a retail intelligence platform built on the
Online Retail II dataset.
The state of the project at the end of each chapter is committed under a
matching git tag (chapter-01, chapter-02, ...).
| Crate | Kind | Role |
|---|---|---|
data-pipeline |
lib | Data engineering (Polars / Arrow) |
classical-ml |
lib | Regression, classification, SVM/NB/k-NN |
deep-learning |
lib | Neural networks, Burn, computer vision |
nlp |
lib | NLP and transformer work (Candle) |
model-server |
bin | REST API (axum) |
mlrs-cli |
bin | Orchestrates each chapter's pipeline |
Pinned to Rust 1.85.0 via rust-toolchain.toml.
Workspace skeleton with all six crates compiling. The data-pipeline crate
holds the first functional module: TransactionRecord, PipelineError, and
the parallel normalize_prices utility with unit tests.
cargo test -p data-pipeline
cargo run -p mlrs-cliDownload online_retail_data.zip, extract, and place the two CSVs in
data/raw/ (kept out of git):
data/raw/online_retail_2009_2010.csv
data/raw/online_retail_2010_2011.csv