ngboost-rs 1.1.1

Natural Gradient Boosting for Probabilistic Prediction - A Rust implementation of NGBoost
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
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246

#[cfg(feature = "accelerate")]
extern crate accelerate_src;

// Serialization tests matching Python's test_pickling.py
//
// Tests that models can be saved and loaded correctly with predictions matching.

use ndarray::{Array1, Array2};
use ndarray_rand::rand_distr::Uniform;
use ndarray_rand::RandomExt;
use ngboost_rs::ngboost::{NGBClassifier, NGBRegressor};
use tempfile::tempdir;

// ============================================================================
// Helper functions
// ============================================================================

fn generate_regression_data(n_samples: usize, n_features: usize) -> (Array2<f64>, Array1<f64>) {
    let x = Array2::random((n_samples, n_features), Uniform::new(0.0, 1.0).unwrap());
    let y = x.column(0).mapv(|v| v * 2.0 + 1.0)
        + x.column(1).mapv(|v| v * 0.5)
        + Array1::random(n_samples, Uniform::new(-0.1, 0.1).unwrap());
    (x, y.mapv(|v| v.abs() + 0.1))
}

fn generate_classification_data(n_samples: usize, n_features: usize) -> (Array2<f64>, Array1<f64>) {
    let x = Array2::random((n_samples, n_features), Uniform::new(0.0, 1.0).unwrap());
    let linear = x.column(0).mapv(|v| v * 2.0) - x.column(1).mapv(|v| v * 1.5);
    let y = linear.mapv(|v| if v > 0.5 { 1.0 } else { 0.0 });
    (x, y)
}

fn arrays_approx_equal(a: &Array1<f64>, b: &Array1<f64>, tol: f64) -> bool {
    if a.len() != b.len() {
        return false;
    }
    a.iter().zip(b.iter()).all(|(x, y)| (x - y).abs() < tol)
}

// ============================================================================
// Regression serialization tests
// ============================================================================

#[test]
fn test_regressor_save_load() {
    let (x, y) = generate_regression_data(100, 5);

    // Train model
    let mut model = NGBRegressor::new(50, 0.1);
    model.fit(&x, &y).expect("Fit should succeed");

    // Get predictions before saving
    let preds_before = model.predict(&x);

    // Create temp directory and save
    let dir = tempdir().expect("Failed to create temp dir");
    let path = dir.path().join("model.bin");
    let path_str = path.to_str().unwrap();

    model.save_model(path_str).expect("Save should succeed");

    // Verify file exists
    assert!(path.exists(), "Model file should exist");

    // Load model
    let loaded_model = NGBRegressor::load_model(path_str).expect("Load should succeed");

    // Get predictions after loading
    let preds_after = loaded_model.predict(&x);

    // Predictions should be identical
    assert!(
        arrays_approx_equal(&preds_before, &preds_after, 1e-10),
        "Predictions should match after load"
    );

    // Cleanup
    dir.close().expect("Failed to close temp dir");
}

#[test]
fn test_classifier_save_load() {
    let (x, y) = generate_classification_data(100, 5);

    // Train model
    let mut model = NGBClassifier::new(30, 0.1);
    model.fit(&x, &y).expect("Fit should succeed");

    // Get predictions before saving
    let preds_before = model.predict(&x);
    let proba_before = model.predict_proba(&x);

    // Create temp directory and save
    let dir = tempdir().expect("Failed to create temp dir");
    let path = dir.path().join("classifier.bin");
    let path_str = path.to_str().unwrap();

    model.save_model(path_str).expect("Save should succeed");

    // Load model
    let loaded_model = NGBClassifier::load_model(path_str).expect("Load should succeed");

    // Get predictions after loading
    let preds_after = loaded_model.predict(&x);
    let proba_after = loaded_model.predict_proba(&x);

    // Predictions should be identical
    assert!(
        arrays_approx_equal(&preds_before, &preds_after, 1e-10),
        "Class predictions should match after load"
    );

    // Probabilities should be identical
    for i in 0..proba_before.nrows() {
        for j in 0..proba_before.ncols() {
            let diff = (proba_before[[i, j]] - proba_after[[i, j]]).abs();
            assert!(diff < 1e-10, "Probabilities should match after load");
        }
    }

    // Cleanup
    dir.close().expect("Failed to close temp dir");
}

#[test]
fn test_regressor_save_load_with_options() {
    let (x, y) = generate_regression_data(100, 5);

    // Train model with various options
    let mut model = NGBRegressor::with_options(
        100, 0.05, true, 0.8, // minibatch_frac
        0.9, // col_sample
        false, 50.0, 1e-5, None, 0.15, false,
    );
    model.fit(&x, &y).expect("Fit should succeed");

    let preds_before = model.predict(&x);

    let dir = tempdir().expect("Failed to create temp dir");
    let path = dir.path().join("model_options.bin");
    let path_str = path.to_str().unwrap();

    model.save_model(path_str).expect("Save should succeed");
    let loaded_model = NGBRegressor::load_model(path_str).expect("Load should succeed");

    let preds_after = loaded_model.predict(&x);

    assert!(
        arrays_approx_equal(&preds_before, &preds_after, 1e-10),
        "Predictions should match for model with options"
    );

    dir.close().expect("Failed to close temp dir");
}

#[test]
fn test_save_load_new_data() {
    let (x_train, y_train) = generate_regression_data(100, 5);
    let (x_test, _) = generate_regression_data(20, 5);

    // Train model
    let mut model = NGBRegressor::new(50, 0.1);
    model.fit(&x_train, &y_train).expect("Fit should succeed");

    // Get predictions on test data
    let preds_before = model.predict(&x_test);

    // Save and load
    let dir = tempdir().expect("Failed to create temp dir");
    let path = dir.path().join("model.bin");
    let path_str = path.to_str().unwrap();

    model.save_model(path_str).expect("Save should succeed");
    let loaded_model = NGBRegressor::load_model(path_str).expect("Load should succeed");

    // Predictions on test data should match
    let preds_after = loaded_model.predict(&x_test);

    assert!(
        arrays_approx_equal(&preds_before, &preds_after, 1e-10),
        "Test predictions should match after load"
    );

    dir.close().expect("Failed to close temp dir");
}

#[test]
fn test_load_nonexistent_file() {
    let result = NGBRegressor::load_model("/nonexistent/path/model.bin");
    assert!(result.is_err(), "Loading nonexistent file should fail");
}

#[test]
fn test_multiple_save_load_cycles() {
    let (x, y) = generate_regression_data(80, 4);

    let mut model = NGBRegressor::new(30, 0.1);
    model.fit(&x, &y).expect("Fit should succeed");

    let original_preds = model.predict(&x);

    let dir = tempdir().expect("Failed to create temp dir");

    // Save/load multiple times
    for i in 0..3 {
        let path = dir.path().join(format!("model_{}.bin", i));
        let path_str = path.to_str().unwrap();

        model.save_model(path_str).expect("Save should succeed");
        let loaded = NGBRegressor::load_model(path_str).expect("Load should succeed");

        let preds = loaded.predict(&x);
        assert!(
            arrays_approx_equal(&original_preds, &preds, 1e-10),
            "Predictions should remain consistent through save/load cycles"
        );
    }

    dir.close().expect("Failed to close temp dir");
}

#[test]
fn test_save_creates_file() {
    let (x, y) = generate_regression_data(50, 3);

    let mut model = NGBRegressor::new(20, 0.1);
    model.fit(&x, &y).expect("Fit should succeed");

    let dir = tempdir().expect("Failed to create temp dir");
    let path = dir.path().join("test_model.bin");
    let path_str = path.to_str().unwrap();

    // File should not exist before save
    assert!(!path.exists());

    model.save_model(path_str).expect("Save should succeed");

    // File should exist after save
    assert!(path.exists());

    // File should have content
    let metadata = std::fs::metadata(&path).expect("Should get metadata");
    assert!(metadata.len() > 0, "File should not be empty");

    dir.close().expect("Failed to close temp dir");
}