dataset_ml/

banknote_authentication.rs

//! Banknote Authentication dataset.
//!
//! The dataset holds features extracted from images of genuine and forged
//! banknote-like specimens. Researchers digitized the images with an
//! industrial camera normally used for print inspection. They then used a
//! Wavelet Transform tool to derive four continuous statistics from each
//! image. The task is to predict the class of a specimen from those four
//! features.
//!
//! **Features (4, all numeric):** `variance`, `skewness`, `curtosis`, and
//! `entropy` of the Wavelet-Transformed image, all continuous `f64` values.
//!
//! **Target:** `class`, the raw integer code from the source, `0` or `1`
//!
//! **Samples:** 1372 total (762 of class `0`, 610 of class `1`)
//! **Application:** Binary classification / banknote authentication
//!
//! **Source:** UCI Machine Learning Repository
//! <https://doi.org/10.24432/C55P57>

use crate::DOWNLOAD_RETRIES;
use crate::traits::impl_ml_dataset;
use dataset_core::{Dataset, DatasetError, acquire_dataset, download_to_with_retries, unzip};
use ndarray::{Array1, Array2};
use std::fs::File;

use csv::ReaderBuilder;

/// The URL for the Banknote Authentication dataset.
///
/// This is the UCI static package. It is a ZIP archive that contains a single
/// file, `data_banknote_authentication.txt`.
///
/// # Citation
///
/// V. Lohweg. "Banknote Authentication," UCI Machine Learning Repository,
/// \[Online\]. Available: <https://doi.org/10.24432/C55P57>
const BANKNOTE_AUTHENTICATION_DATA_URL: &str =
    "https://archive.ics.uci.edu/static/public/267/banknote+authentication.zip";

/// The name the downloaded ZIP archive is saved under inside the temp directory.
const BANKNOTE_AUTHENTICATION_ZIP_FILENAME: &str = "banknote_authentication.zip";

/// The name of the only file inside the archive, holding all 1372 records.
const BANKNOTE_AUTHENTICATION_SOURCE_FILENAME: &str = "data_banknote_authentication.txt";

/// The name of the final cached Banknote Authentication dataset file.
const BANKNOTE_AUTHENTICATION_FILENAME: &str = "banknote_authentication.csv";

/// The SHA256 hash of the Banknote Authentication dataset file
/// (`data_banknote_authentication.txt`).
const BANKNOTE_AUTHENTICATION_SHA256: &str =
    "d0539aaed2139ba7a587b3e34fb345ce503ff7d5d33dbf9912d8e195ce425cb9";

/// The name of the dataset.
const BANKNOTE_AUTHENTICATION_DATASET_NAME: &str = "banknote_authentication";

/// Number of samples.
const N_SAMPLES: usize = 1372;

/// The number of numeric features per sample.
const N_FEATURES: usize = 4;

/// The number of columns per CSV record (4 features + 1 label).
const N_COLUMNS: usize = N_FEATURES + 1;

/// The names of the four feature columns, in source order. `curtosis` keeps the
/// (misspelled) UCI attribute name so the schema matches the source exactly.
const FEATURE_NAMES: [&str; N_FEATURES] = ["variance", "skewness", "curtosis", "entropy"];

/// Type alias for the Banknote Authentication dataset: (features, labels).
type BanknoteAuthenticationData = (Array2<f64>, Array1<u8>);

/// This struct represents the Banknote Authentication dataset and loads it lazily.
///
/// Nothing loads until you call a data accessor method. After loading, the
/// data stays cached for later accesses.
///
/// # About Dataset
///
/// Researchers extracted the data from images of genuine and forged
/// banknote-like specimens. They digitized the images with an industrial
/// camera normally used for print inspection. This camera produced 400×400
/// pixel grayscale images at a resolution of about 660 dpi. Researchers then
/// used a Wavelet Transform tool to extract four continuous statistics from
/// each image. These statistics are the variance, skewness, curtosis, and
/// entropy of the transformed image. Together they form a compact,
/// pure-numeric feature matrix over 1372 specimens.
///
/// # Feature columns
///
/// All 4 features are quantitative, stored in one `(1372, 4)` `Array2<f64>`
/// matrix. By 0-based column index:
///
/// | Column | Attribute  | Description                                          |
/// |--------|------------|------------------------------------------------------|
/// | `0`    | `variance` | variance of the Wavelet-Transformed image            |
/// | `1`    | `skewness` | skewness of the Wavelet-Transformed image            |
/// | `2`    | `curtosis` | curtosis of the Wavelet-Transformed image            |
/// | `3`    | `entropy`  | entropy of the image                                 |
///
/// `curtosis` keeps the source's spelling (UCI names the attribute that way)
/// so the schema matches the source exactly.
///
/// # Labels
///
/// - `class` (shape `(1372,)`): the `Array1<u8>` holds the raw integer code from
///   the source (`0` or `1`). UCI does not document which code corresponds to
///   genuine vs forged notes, so the loader exposes it verbatim.
///
/// See more information at
/// <https://archive.ics.uci.edu/dataset/267/banknote+authentication>.
///
/// # Citation
///
/// V. Lohweg. "Banknote Authentication," UCI Machine Learning Repository,
/// \[Online\]. Available: <https://doi.org/10.24432/C55P57>
///
/// # Thread Safety
///
/// Every field implements `Send` and `Sync`, so this struct implements them too. It is safe
/// to share across threads.
/// The internal [`Dataset`] makes initialization thread-safe and lazy.
///
/// # Example
/// ```no_run
/// use dataset_ml::banknote_authentication::BanknoteAuthentication;
///
/// let download_dir = "./banknote_authentication"; // creates the directory if it is missing
///
/// let mut dataset = BanknoteAuthentication::new(download_dir);
/// let features = dataset.features().unwrap();
/// let labels = dataset.labels().unwrap();
///
/// let (features, labels) = dataset.data().unwrap(); // also a way to get features and labels
/// assert_eq!(features.shape(), &[1372, 4]);
/// assert_eq!(labels.len(), 1372);
///
/// // `get_data()` borrows the cached arrays without reloading. `get_data_mut()`
/// // edits them in place. It needs no clone and no reload, and the change
/// // stays cached. Prefer this method over cloning with `.to_owned()` when
/// // you only need to change values.
/// if let Some((features, labels)) = dataset.get_data_mut() {
///     features[[0, 0]] = 0.5;
///     labels[0] = 1;
/// }
/// assert!(dataset.get_data().is_some());
///
/// // `take_data()` moves owned arrays out (no `to_owned()` clone) and leaves the
/// // instance reusable. The next access reloads from the cached file.
/// let (owned_features, owned_labels) = dataset.take_data().unwrap();
/// assert_eq!(owned_features.shape(), &[1372, 4]);
/// assert_eq!(owned_labels.len(), 1372);
///
/// // `into_data()` also returns owned arrays with no clone, but consumes the
/// // instance (use it when you are done with the dataset).
/// let (owned_features, owned_labels) = dataset.into_data().unwrap();
/// assert_eq!(owned_features.shape(), &[1372, 4]);
/// assert_eq!(owned_labels.len(), 1372);
/// ```
#[derive(Debug)]
pub struct BanknoteAuthentication {
    dataset: Dataset<BanknoteAuthenticationData, DatasetError>,
}

impl BanknoteAuthentication {
    /// Create a new BanknoteAuthentication instance without loading data.
    ///
    /// This does not load the dataset. The dataset loads on the first call to a
    /// data accessor method. This is a lightweight operation: it only stores the
    /// storage directory.
    ///
    /// # Parameters
    ///
    /// - `storage_dir` - Directory used to store the dataset.
    ///
    /// # Returns
    ///
    /// - `Self` - `BanknoteAuthentication` instance ready for lazy loading.
    pub fn new(storage_dir: &str) -> Self {
        BanknoteAuthentication {
            dataset: Dataset::new(storage_dir, Self::load_data),
        }
    }

    /// Get and parse the Banknote Authentication dataset.
    fn load_data(dir: &str) -> Result<BanknoteAuthenticationData, DatasetError> {
        // Prepare the dataset file: download the UCI ZIP package, extract it, and
        // surface the single `data_banknote_authentication.txt` file it contains.
        let file_path = acquire_dataset(
            dir,
            BANKNOTE_AUTHENTICATION_FILENAME,
            BANKNOTE_AUTHENTICATION_DATASET_NAME,
            Some(BANKNOTE_AUTHENTICATION_SHA256),
            |temp_path| {
                download_to_with_retries(
                    BANKNOTE_AUTHENTICATION_DATA_URL,
                    temp_path,
                    Some(BANKNOTE_AUTHENTICATION_ZIP_FILENAME),
                    DOWNLOAD_RETRIES,
                )?;
                unzip(
                    &temp_path.join(BANKNOTE_AUTHENTICATION_ZIP_FILENAME),
                    temp_path,
                )?;
                Ok(temp_path.join(BANKNOTE_AUTHENTICATION_SOURCE_FILENAME))
            },
        )?;

        // The source is plain comma-separated with no header: every line is a
        // record of 4 numeric features followed by the class code.
        let file = File::open(&file_path)?;
        let mut rdr = ReaderBuilder::new().has_headers(false).from_reader(file);

        let mut features: Vec<f64> = Vec::with_capacity(N_SAMPLES * N_FEATURES);
        let mut labels: Vec<u8> = Vec::with_capacity(N_SAMPLES);

        for (idx, result) in rdr.records().enumerate() {
            let record = result.map_err(|e| {
                DatasetError::csv_read_error(BANKNOTE_AUTHENTICATION_DATASET_NAME, e)
            })?;
            let line_num = idx + 1; // headerless file, lines are 1-indexed

            // Skip blank lines defensively (e.g. a trailing newline).
            if record.iter().all(|f| f.is_empty()) {
                continue;
            }

            if record.len() != N_COLUMNS {
                return Err(DatasetError::invalid_column_count(
                    BANKNOTE_AUTHENTICATION_DATASET_NAME,
                    N_COLUMNS,
                    record.len(),
                    line_num,
                ));
            }

            // 4 numeric features.
            for (col, name) in FEATURE_NAMES.iter().enumerate() {
                let value: f64 = record[col].trim().parse().map_err(|e| {
                    DatasetError::parse_failed(
                        BANKNOTE_AUTHENTICATION_DATASET_NAME,
                        name,
                        line_num,
                        e,
                    )
                })?;
                features.push(value);
            }

            // Label, kept as the raw `0`/`1` code the source records.
            let raw_label = record[N_FEATURES].trim();
            let label: u8 = raw_label.parse().map_err(|e| {
                DatasetError::parse_failed(
                    BANKNOTE_AUTHENTICATION_DATASET_NAME,
                    "class",
                    line_num,
                    e,
                )
            })?;
            if label > 1 {
                return Err(DatasetError::invalid_value(
                    BANKNOTE_AUTHENTICATION_DATASET_NAME,
                    "class",
                    raw_label,
                    line_num,
                ));
            }
            labels.push(label);
        }

        let n_samples = labels.len();
        if n_samples == 0 {
            return Err(DatasetError::empty_dataset(
                BANKNOTE_AUTHENTICATION_DATASET_NAME,
            ));
        }

        // Banknote Authentication has a fixed schema of 4 numeric features per sample.
        let features_array =
            Array2::from_shape_vec((n_samples, N_FEATURES), features).map_err(|e| {
                DatasetError::array_shape_error(BANKNOTE_AUTHENTICATION_DATASET_NAME, "features", e)
            })?;

        let labels_array = Array1::from_vec(labels);

        Ok((features_array, labels_array))
    }

    /// Get a reference to the feature matrix.
    ///
    /// This method triggers lazy loading on first call. Later calls return the
    /// cached data instantly.
    ///
    /// # Returns
    ///
    /// - `&Array2<f64>` - Reference to the numeric feature matrix with shape
    ///   `(1372, 4)`: the `variance`, `skewness`, `curtosis`, and `entropy` of
    ///   each Wavelet-Transformed image.
    ///
    /// # Errors
    ///
    /// Returns `DatasetError` if:
    /// - Download fails due to network issues
    /// - File extraction or I/O operations fail
    /// - Data format is invalid (wrong number of columns, unparseable values, or invalid labels)
    /// - Dataset size does not match the expected dimensions (1372 samples, 4 features)
    pub fn features(&self) -> Result<&Array2<f64>, DatasetError> {
        Ok(&self.dataset.load()?.0)
    }

    /// Get a reference to the labels vector.
    ///
    /// This method triggers lazy loading on first call. Later calls return the
    /// cached data instantly.
    ///
    /// # Returns
    ///
    /// - `&Array1<u8>` - Reference to labels vector with shape `(1372,)`
    ///   containing the raw class codes (`0` or `1`).
    ///
    /// # Errors
    ///
    /// Returns `DatasetError` if:
    /// - Download fails due to network issues
    /// - File extraction or I/O operations fail
    /// - Data format is invalid (wrong number of columns, unparseable values, or invalid labels)
    /// - Dataset size does not match the expected dimensions (1372 samples)
    pub fn labels(&self) -> Result<&Array1<u8>, DatasetError> {
        Ok(&self.dataset.load()?.1)
    }

    /// Get both features and labels as references.
    ///
    /// This method triggers lazy loading on first call. Later calls return the
    /// cached data instantly.
    ///
    /// # Returns
    ///
    /// - `&BanknoteAuthenticationData` - reference to the cached
    ///   `(features, labels)` tuple: the feature matrix has shape `(1372, 4)` and
    ///   the label vector has shape `(1372,)` containing the raw class codes
    ///   (`0` or `1`).
    ///
    /// # Errors
    ///
    /// Returns `DatasetError` if:
    /// - Download fails due to network issues
    /// - File extraction or I/O operations fail
    /// - Data format is invalid (wrong number of columns, unparseable values, or invalid labels)
    /// - Dataset size does not match the expected dimensions (1372 samples, 4 features)
    pub fn data(&self) -> Result<&BanknoteAuthenticationData, DatasetError> {
        self.dataset.load()
    }

    /// Get both features and labels as references, without triggering loading.
    ///
    /// Unlike [`BanknoteAuthentication::data`], which loads the dataset on first
    /// call, this never runs the loader. If the data has not been loaded yet, it
    /// returns `None` instead of downloading and parsing.
    ///
    /// Use this method when you want the data only if it is already cached. This
    /// avoids the download and parse cost when the data is not cached.
    ///
    /// # Returns
    ///
    /// - `Some(&BanknoteAuthenticationData)` - reference to the cached
    ///   `(features, labels)` tuple (feature matrix `(1372, 4)`, label vector
    ///   `(1372,)`), if loaded.
    /// - `None` - if the dataset has not been loaded yet.
    pub fn get_data(&self) -> Option<&BanknoteAuthenticationData> {
        self.dataset.get()
    }

    /// Get mutable references to features and labels for **in-place** editing.
    ///
    /// This lets you change the cached arrays directly (e.g. normalize features,
    /// replace label values). It needs no `to_owned()` clone, and the arrays
    /// stay in the cache. The changes persist, so later calls to
    /// [`BanknoteAuthentication::features`], [`BanknoteAuthentication::data`], or
    /// [`BanknoteAuthentication::get_data`] see them.
    ///
    /// Like [`BanknoteAuthentication::get_data`], this does **not** trigger
    /// loading. It returns `None` if the dataset has not been loaded. If you
    /// need to make sure the data is present, call a loading accessor first
    /// (e.g. [`BanknoteAuthentication::data`]).
    ///
    /// # Returns
    ///
    /// - `Some(&mut BanknoteAuthenticationData)` - mutable reference to the cached
    ///   `(features, labels)` tuple (feature matrix `(1372, 4)`, label vector
    ///   `(1372,)`), if loaded.
    /// - `None` - if the dataset has not been loaded yet.
    pub fn get_data_mut(&mut self) -> Option<&mut BanknoteAuthenticationData> {
        self.dataset.get_mut()
    }

    /// Consume the dataset and return **owned** features and labels.
    ///
    /// Unlike [`BanknoteAuthentication::data`], which borrows the cached data,
    /// this moves it out and returns owned arrays directly. It needs no
    /// `to_owned()` clone. The dataset is loaded on first access if it has not
    /// been loaded yet.
    ///
    /// This **consumes** `self`, so the instance cannot be used afterwards. If you
    /// want owned data but need to keep using the instance, use
    /// [`BanknoteAuthentication::take_data`] instead. It takes `&mut self` and
    /// leaves the instance reusable.
    ///
    /// # Returns
    ///
    /// - `(Array2<f64>, Array1<u8>)` - owned feature matrix with shape
    ///   `(1372, 4)` and owned label vector with shape `(1372,)`.
    ///
    /// # Errors
    ///
    /// Returns `DatasetError` if loading fails (network, file I/O, parsing, invalid
    /// labels, or a dimension mismatch).
    pub fn into_data(self) -> Result<BanknoteAuthenticationData, DatasetError> {
        self.dataset.load()?;
        Ok(self
            .dataset
            .into_inner()
            .expect("data is present after a successful load"))
    }

    /// Take **owned** features and labels out of the dataset. The instance stays
    /// reusable.
    ///
    /// Like [`BanknoteAuthentication::into_data`], this returns owned arrays with
    /// no `to_owned()` clone. But instead of consuming the instance, it takes
    /// `&mut self` and moves the cached data out. This resets the instance to
    /// its unloaded state. The next accessor call (e.g.
    /// [`BanknoteAuthentication::features`] or [`BanknoteAuthentication::data`])
    /// loads the dataset again.
    ///
    /// If you are done with the instance, use
    /// [`BanknoteAuthentication::into_data`] instead.
    ///
    /// # Returns
    ///
    /// - `(Array2<f64>, Array1<u8>)` - owned feature matrix with shape
    ///   `(1372, 4)` and owned label vector with shape `(1372,)`.
    ///
    /// # Errors
    ///
    /// Returns `DatasetError` if loading fails (network, file I/O, parsing, invalid
    /// labels, or a dimension mismatch).
    pub fn take_data(&mut self) -> Result<BanknoteAuthenticationData, DatasetError> {
        self.dataset.load()?;
        Ok(self
            .dataset
            .take()
            .expect("data is present after a successful load"))
    }
}

impl_ml_dataset!(
    BanknoteAuthentication,
    BanknoteAuthenticationData,
    "banknote_authentication"
);