dataset_core/datasets/
iris.rs

1use crate::{Dataset, DatasetError, acquire_dataset, download_to};
2use csv::ReaderBuilder;
3use ndarray::{Array1, Array2};
4use std::fs::File;
5
6/// The URL for the Iris dataset.
7///
8/// # Citation
9///
10/// R. A. Fisher. "Iris," UCI Machine Learning Repository, \[Online\].
11/// Available: <https://doi.org/10.24432/C56C76>
12const IRIS_DATA_URL: &str = "https://gist.githubusercontent.com/curran/a08a1080b88344b0c8a7/raw/0e7a9b0a5d22642a06d3d5b9bcbad9890c8ee534/iris.csv";
13
14/// The name of the Iris dataset file.
15const IRIS_FILENAME: &str = "iris.csv";
16
17/// The SHA256 hash of the Iris dataset file.
18const IRIS_SHA256: &str = "c52742e50315a99f956a383faedf7575552675f6409ef0f9a47076dd08479930";
19
20/// The name of the dataset
21const IRIS_DATASET_NAME: &str = "iris";
22
23/// A struct representing the Iris dataset with lazy loading.
24///
25/// The dataset is not loaded until you call one of the data accessor methods.
26/// Once loaded, the data is cached for subsequent accesses.
27///
28/// # About Dataset
29///
30/// The Iris dataset is a classic dataset for classification tasks. It includes three iris species
31/// with 50 samples each as well as some properties about each flower. One flower species is
32/// linearly separable from the other two, but the other two are not linearly separable from each other.
33///
34/// Features:
35/// - sepal length in cm
36/// - sepal width in cm
37/// - petal length in cm
38/// - petal width in cm
39///
40/// Labels:
41/// - species name (in `&str`): `"setosa"`, `"versicolor"`, `"virginica"`
42///
43/// See more information at <https://archive.ics.uci.edu/dataset/53/iris>
44///
45/// # Citation
46///
47/// R. A. Fisher. "Iris," UCI Machine Learning Repository, \[Online\].
48/// Available: <https://doi.org/10.24432/C56C76>
49///
50/// # Thread Safety
51///
52/// This struct automatically implements `Send` and `Sync` (All fields implement them), making it safe to share across threads.
53/// The internal [`Dataset`] ensures thread-safe lazy initialization.
54///
55/// # Example
56/// ```rust
57/// use dataset_core::datasets::iris::Iris;
58///
59/// let download_dir = "./iris"; // the code will create the directory if it doesn't exist
60///
61/// let dataset = Iris::new(download_dir);
62/// let features = dataset.features().unwrap();
63/// let labels = dataset.labels().unwrap();
64///
65/// let (features, labels) = dataset.data().unwrap(); // this is also a way to get features and labels
66/// // you can use `.to_owned()` to get owned copies of the data
67/// let mut features_owned = features.to_owned();
68/// let mut labels_owned = labels.to_owned();
69///
70/// // Example: Modify feature values
71/// features_owned[[0, 0]] = 5.5;
72/// labels_owned[0] = "setosa-modified";
73///
74/// assert_eq!(features.shape(), &[150, 4]);
75/// assert_eq!(labels.len(), 150);
76///
77/// // clean up: remove the downloaded files (dispensable)
78/// std::fs::remove_dir_all(download_dir).unwrap();
79/// ```
80#[derive(Debug)]
81pub struct Iris {
82    dataset: Dataset<(Array2<f64>, Array1<&'static str>)>,
83}
84
85impl Iris {
86    /// Create a new Iris instance without loading data.
87    ///
88    /// The dataset will be loaded lazily when you first call any data accessor method.
89    /// This is a lightweight operation that only stores the storage directory.
90    ///
91    /// # Parameters
92    ///
93    /// - `storage_dir` - Directory where the dataset will be stored.
94    ///
95    /// # Returns
96    ///
97    /// - `Self` - `Iris` instance ready for lazy loading.
98    pub fn new(storage_dir: &str) -> Self {
99        Iris {
100            dataset: Dataset::new(storage_dir),
101        }
102    }
103
104    /// Acquire and parse the Iris dataset.
105    fn load_data(dir: &str) -> Result<(Array2<f64>, Array1<&'static str>), DatasetError> {
106        // Prepare the dataset file
107        let file_path = acquire_dataset(
108            dir,
109            IRIS_FILENAME,
110            IRIS_DATASET_NAME,
111            Some(IRIS_SHA256),
112            |temp_path| {
113                download_to(IRIS_DATA_URL, temp_path, None)?;
114                Ok(temp_path.join(IRIS_FILENAME))
115            },
116        )?;
117
118        // Parse the file
119        let file = File::open(&file_path)?;
120        let mut rdr = ReaderBuilder::new().has_headers(true).from_reader(file);
121
122        let mut features = Vec::new();
123        let mut labels = Vec::new();
124        let mut num_features: Option<usize> = None;
125
126        for (idx, result) in rdr.records().enumerate() {
127            let record = result.map_err(|e| DatasetError::csv_read_error(IRIS_DATASET_NAME, e))?;
128            let line_num = idx + 2; // +1 for 0-indexed, +1 for header
129
130            if num_features.is_none() {
131                if record.len() < 2 {
132                    return Err(DatasetError::invalid_column_count(
133                        IRIS_DATASET_NAME,
134                        2,
135                        record.len(),
136                        line_num,
137                        &format!("{:?}", record),
138                    ));
139                }
140                num_features = Some(record.len() - 1);
141            }
142
143            let n_features = num_features.unwrap();
144            if record.len() != n_features + 1 {
145                return Err(DatasetError::invalid_column_count(
146                    IRIS_DATASET_NAME,
147                    n_features + 1,
148                    record.len(),
149                    line_num,
150                    &format!("{:?}", record),
151                ));
152            }
153
154            for i in 0..n_features {
155                features.push(record[i].parse::<f64>().map_err(|e| {
156                    let field = format!("feature[{i}]");
157                    DatasetError::parse_failed(
158                        IRIS_DATASET_NAME,
159                        &field,
160                        line_num,
161                        &format!("{:?}", record),
162                        e,
163                    )
164                })?);
165            }
166
167            labels.push(match &record[n_features] {
168                "setosa" => "setosa",
169                "versicolor" => "versicolor",
170                "virginica" => "virginica",
171                other => {
172                    return Err(DatasetError::invalid_value(
173                        IRIS_DATASET_NAME,
174                        "label",
175                        other,
176                        line_num,
177                        &format!("{:?}", record),
178                    ));
179                }
180            });
181        }
182
183        let n_samples = labels.len();
184        if n_samples == 0 {
185            return Err(DatasetError::empty_dataset(IRIS_DATASET_NAME));
186        }
187
188        let n_features = num_features.unwrap();
189        let features_array = Array2::from_shape_vec((n_samples, n_features), features)
190            .map_err(|e| DatasetError::array_shape_error(IRIS_DATASET_NAME, "features", e))?;
191        let labels_array = Array1::from_vec(labels);
192
193        Ok((features_array, labels_array))
194    }
195
196    /// Get a reference to the feature matrix.
197    ///
198    /// This method triggers lazy loading on first call. Subsequent calls return
199    /// the cached data instantly.
200    ///
201    /// # Returns
202    ///
203    /// - `&Array2<f64>` - Reference to feature matrix with shape `(150, 4)` containing:
204    ///     - sepal length in cm
205    ///     - sepal width in cm
206    ///     - petal length in cm
207    ///     - petal width in cm
208    ///
209    /// # Errors
210    ///
211    /// Returns `DatasetError` if:
212    /// - Download fails due to network issues
213    /// - File extraction or I/O operations fail
214    /// - Data format is invalid (wrong number of columns, unparseable values, or invalid labels)
215    /// - Dataset size doesn't match expected dimensions (150 samples, 4 features)
216    pub fn features(&self) -> Result<&Array2<f64>, DatasetError> {
217        Ok(&self.dataset.load(Self::load_data)?.0)
218    }
219
220    /// Get a reference to the labels vector.
221    ///
222    /// This method triggers lazy loading on first call. Subsequent calls return
223    /// the cached data instantly.
224    ///
225    /// # Returns
226    ///
227    /// - `&Array1<&'static str>` - Reference to labels vector with shape `(150,)` containing species names (`"setosa"`, `"versicolor"`, `"virginica"`)
228    ///
229    /// # Errors
230    ///
231    /// Returns `DatasetError` if:
232    /// - Download fails due to network issues
233    /// - File extraction or I/O operations fail
234    /// - Data format is invalid (wrong number of columns, unparseable values, or invalid labels)
235    /// - Dataset size doesn't match expected dimensions (150 samples)
236    pub fn labels(&self) -> Result<&Array1<&'static str>, DatasetError> {
237        Ok(&self.dataset.load(Self::load_data)?.1)
238    }
239
240    /// Get both features and labels as references.
241    ///
242    /// This method triggers lazy loading on first call. Subsequent calls return
243    /// the cached data instantly.
244    ///
245    /// # Returns
246    ///
247    /// - `&Array2<f64>` - Reference to feature matrix with shape `(150, 4)` containing:
248    ///     - sepal length in cm
249    ///     - sepal width in cm
250    ///     - petal length in cm
251    ///     - petal width in cm
252    /// - `&Array1<&'static str>` - Reference to labels vector with shape `(150,)` containing species names (`"setosa"`, `"versicolor"`, `"virginica"`)
253    ///
254    /// # Errors
255    ///
256    /// Returns `DatasetError` if:
257    /// - Download fails due to network issues
258    /// - File extraction or I/O operations fail
259    /// - Data format is invalid (wrong number of columns, unparseable values, or invalid labels)
260    /// - Dataset size doesn't match expected dimensions (150 samples, 4 features)
261    pub fn data(&self) -> Result<(&Array2<f64>, &Array1<&'static str>), DatasetError> {
262        let data = self.dataset.load(Self::load_data)?;
263        Ok((&data.0, &data.1))
264    }
265}
dataset_core/datasets/iris.rs

dataset_core/datasets/
iris.rs
