rustyml 0.11.0

A high-performance machine learning & deep learning library in pure Rust, offering ML algorithms and neural network support
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167

#![cfg(feature = "utility")]

use ndarray::*;
use rustyml::utility::linear_discriminant_analysis::*;
use std::error::Error;

fn make_three_class_dataset() -> (Array2<f64>, Array1<i32>) {
    let x = arr2(&[
        [0.0, 0.2],
        [0.2, -0.1],
        [-0.1, 0.1],
        [5.0, 5.1],
        [5.2, 4.8],
        [4.8, 5.0],
        [10.0, 0.2],
        [9.9, -0.1],
        [10.1, 0.1],
    ]);
    let y = Array1::from_vec(vec![0, 0, 0, 1, 1, 1, 2, 2, 2]);
    (x, y)
}

fn classification_accuracy(pred: &Array1<i32>, y: &Array1<i32>) -> f64 {
    let correct = pred.iter().zip(y.iter()).filter(|(a, b)| a == b).count();
    correct as f64 / y.len() as f64
}

#[test]
fn test_lda_new_and_default() {
    let lda = LDA::new(2, None, None).unwrap();
    assert_eq!(lda.get_n_components(), 2);
    assert_eq!(lda.get_solver(), Solver::SVD);
    assert!(lda.get_shrinkage().is_none());
    assert!(lda.get_classes().is_none());
    assert!(lda.get_priors().is_none());
    assert!(lda.get_means().is_none());
    assert!(lda.get_cov_inv().is_none());
    assert!(lda.get_projection().is_none());

    let default_lda = LDA::default();
    assert_eq!(default_lda.get_n_components(), 2);
    assert_eq!(default_lda.get_solver(), Solver::SVD);
}

#[test]
fn test_lda_new_validation() {
    assert!(LDA::new(0, None, None).is_err());
    assert!(LDA::new(2, None, Some(Shrinkage::Manual(-0.1))).is_err());
    assert!(LDA::new(2, None, Some(Shrinkage::Manual(1.1))).is_err());
    assert!(LDA::new(2, None, Some(Shrinkage::Manual(f64::NAN))).is_err());
    assert!(LDA::new(2, None, Some(Shrinkage::Manual(0.25))).is_ok());
}

#[test]
fn test_fit_predict_transform() -> Result<(), Box<dyn Error>> {
    let (x, y) = make_three_class_dataset();
    let mut lda = LDA::new(2, Some(Solver::SVD), None)?;

    lda.fit(&x.view(), &y.view())?;

    assert!(lda.get_projection().is_some());
    assert_eq!(lda.get_projection().unwrap().shape(), &[2, 2]);

    let predictions = lda.predict(&x.view())?;
    assert_eq!(predictions.len(), y.len());
    assert!(classification_accuracy(&predictions, &y) >= 0.9);

    let transformed = lda.transform(&x.view())?;
    assert_eq!(transformed.shape(), &[x.nrows(), 2]);

    Ok(())
}

#[test]
fn test_fit_transform() -> Result<(), Box<dyn Error>> {
    let (x, y) = make_three_class_dataset();

    let mut lda = LDA::new(2, Some(Solver::SVD), Some(Shrinkage::Auto))?;
    let transformed = lda.fit_transform(&x.view(), &y.view())?;
    assert_eq!(transformed.shape(), &[x.nrows(), 2]);
    assert!(lda.get_projection().is_some());

    Ok(())
}

#[test]
fn test_fit_with_solvers() -> Result<(), Box<dyn Error>> {
    let (x, y) = make_three_class_dataset();
    let solvers = [Solver::SVD, Solver::Eigen, Solver::LSQR];

    for solver in solvers {
        let mut lda = LDA::new(2, Some(solver), None)?;
        lda.fit(&x.view(), &y.view())?;
        let predictions = lda.predict(&x.view())?;
        assert_eq!(predictions.len(), y.len());
        assert!(classification_accuracy(&predictions, &y) >= 0.7);
        let transformed = lda.transform(&x.view())?;
        assert_eq!(transformed.shape(), &[x.nrows(), 2]);
    }

    Ok(())
}

#[test]
fn test_errors_and_validation_paths() {
    let (x, y) = make_three_class_dataset();

    let lda = LDA::new(2, None, None).unwrap();
    assert!(lda.predict(&x.view()).is_err());
    assert!(lda.transform(&x.view()).is_err());

    let mut lda_bad_components = LDA::new(3, None, None).unwrap();
    assert!(lda_bad_components.fit(&x.view(), &y.view()).is_err());

    let mut lda_mismatch = LDA::new(2, None, None).unwrap();
    let y_short = Array1::from_vec(vec![0, 0, 1]);
    assert!(lda_mismatch.fit(&x.view(), &y_short.view()).is_err());

    let x_bad = arr2(&[[1.0, 2.0], [3.0, 4.0]]);
    let y_bad = Array1::from_vec(vec![0, 1]);
    let mut lda_single = LDA::new(1, None, None).unwrap();
    assert!(lda_single.fit(&x_bad.view(), &y_bad.view()).is_err());
}

#[test]
fn test_lda_actual_performance() -> Result<(), Box<dyn Error>> {
    let x_train = arr2(&[
        [-0.2, 0.1],
        [0.1, 0.2],
        [-0.1, -0.2],
        [0.2, -0.1],
        [3.8, 4.1],
        [4.2, 3.9],
        [4.0, 4.2],
        [4.1, 3.8],
        [-0.1, 5.1],
        [0.2, 4.9],
        [0.1, 5.2],
        [-0.2, 4.8],
    ]);
    let y_train = Array1::from_vec(vec![0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2]);

    let x_test = arr2(&[
        [0.0, 0.0],
        [0.15, -0.05],
        [-0.15, 0.05],
        [4.05, 4.05],
        [3.9, 4.0],
        [4.2, 4.1],
        [0.0, 5.0],
        [0.15, 4.95],
        [-0.15, 5.05],
    ]);
    let y_test = Array1::from_vec(vec![0, 0, 0, 1, 1, 1, 2, 2, 2]);

    let mut lda = LDA::new(2, Some(Solver::SVD), Some(Shrinkage::Manual(0.1)))?;
    lda.fit(&x_train.view(), &y_train.view())?;

    let predictions = lda.predict(&x_test.view())?;
    let accuracy = classification_accuracy(&predictions, &y_test);
    assert!(accuracy >= 0.9);

    let transformed = lda.transform(&x_test.view())?;
    assert_eq!(transformed.shape(), &[x_test.nrows(), 2]);

    Ok(())
}