简体中文 | English

dataset-ml

Ready-to-use loaders for classic machine learning datasets, built on dataset-core.

Overview

dataset-ml ships with loaders for ten classic ML datasets. Each loader:

• Downloads the source file on first access (with ureq).
• Verifies a pinned SHA-256 hash to detect corruption or upstream changes.
• Parses the CSV into ndarray Array1 / Array2.
• Caches the parsed result in memory via dataset_core::Dataset<T, E> — subsequent accesses return a & reference with zero I/O.

Each module is also a complete reference implementation of the pattern for wrapping Dataset<T, E> for a concrete data source.

Installation

[dependencies]
dataset-ml = "0.2"

Datasets

All structs are also re-exported at the crate root, so dataset_ml::Iris, dataset_ml::RedWineQuality, etc. work too.

Usage

use dataset_ml::iris::Iris;

fn main() {
    let iris = Iris::new("./data");

    // Lazy: downloads and parses on first access, then cached.
    let features = iris.features().unwrap();  // &Array2<f64>
    let labels   = iris.labels().unwrap();    // &Array1<String>

    // Or get both at once:
    let (features, labels) = iris.data().unwrap();

    assert_eq!(features.shape(), &[150, 4]);
    assert_eq!(labels.len(), 150);

    // Call .to_owned() when you need a mutable copy.
    let mut owned = features.to_owned();
    owned[[0, 0]] = 5.5;
}

Each dataset struct follows the same pattern:

• new(storage_dir) — create instance (no I/O)
• features() — reference to feature matrix
• labels() / targets() — reference to label/target vector
• data() — all references at once

Note: Titanic and Palmer Penguins are mixed-type: features() returns (&Array2<String>, &Array2<f64>) (string + numeric features), and data() returns a triple. Palmer Penguins also represents missing values as NaN (numeric) or "" (string).

Note: California Housing reproduces scikit-learn's fetch_california_housing features by deriving them from the raw census columns (e.g. AveRooms = total_rooms / households) and scaling the target by 1/100000. Its 207 missing total_bedrooms values surface as NaN in AveBedrms.

Migration from dataset-core 0.1.x

If you used the datasets feature of dataset-core 0.1.x, switch to this crate:

- dataset-core = { version = "0.1", features = ["datasets"] }
+ dataset-ml = "0.2"

dataset_core::utils::* and dataset_core::DatasetError are unchanged — they remain in dataset-core under the utils feature.

Performance Considerations

• First access: downloads the file (if not on disk), validates SHA-256, parses, caches in memory.
• Subsequent accesses: return a reference to the cached data — zero allocation, zero I/O.
• .to_owned(): clones cached data into a new owned value — use only when mutation is needed.
• Offline: once downloaded, datasets stay on disk; no network needed on subsequent runs.

License

This project is licensed under the MIT License — see LICENSE for details.

Datasets Attribution

The bundled datasets are classic machine learning datasets widely used for educational and research purposes:

• Iris: Fisher's Iris dataset (1936)
• Breast Cancer Wisconsin (Diagnostic): Wolberg, Mangasarian, Street & Street (1995)
• Boston Housing: Harrison & Rubinfeld (1978)
• California Housing: Pace & Barry (1997), from the 1990 U.S. census
• Diabetes: Pima Indians Diabetes Database
• Titanic: Kaggle Titanic dataset
• Palmer Penguins: Horst, Hill & Gorman (2020); data by Gorman, Williams & Fraser (2014)
• Wine Recognition: Aeberhard & Forina (1991), UCI Machine Learning Repository
• Wine Quality: UCI Machine Learning Repository

Author

SomeB1oody — stanyin64@gmail.com