aprender-core 0.29.2

Next-generation machine learning library in pure Rust
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
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391


    // ========================================================================
    // Gini Impurity Tests
    // ========================================================================

    #[test]
    fn test_gini_impurity_empty() {
        assert!((gini_impurity(&[]) - 0.0).abs() < 1e-7);
    }

    #[test]
    fn test_gini_impurity_pure_single_class() {
        // All same class -> Gini = 0
        let labels = vec![0, 0, 0, 0];
        assert!((gini_impurity(&labels) - 0.0).abs() < 1e-7);
    }

    #[test]
    fn test_gini_impurity_two_classes_balanced() {
        // 50/50 split -> Gini = 0.5
        let labels = vec![0, 1, 0, 1];
        assert!((gini_impurity(&labels) - 0.5).abs() < 1e-7);
    }

    #[test]
    fn test_gini_impurity_two_classes_unbalanced() {
        // 3/4 class 0, 1/4 class 1 -> Gini = 1 - (0.75^2 + 0.25^2) = 1 - 0.625 = 0.375
        let labels = vec![0, 0, 0, 1];
        assert!((gini_impurity(&labels) - 0.375).abs() < 1e-7);
    }

    #[test]
    fn test_gini_impurity_three_classes_uniform() {
        // Three classes equally distributed -> Gini = 1 - 3*(1/3)^2 = 1 - 1/3 = 2/3
        let labels = vec![0, 1, 2, 0, 1, 2];
        let expected = 1.0 - 3.0 * (1.0_f32 / 3.0).powi(2);
        assert!((gini_impurity(&labels) - expected).abs() < 1e-6);
    }

    #[test]
    fn test_gini_impurity_single_sample() {
        let labels = vec![5];
        assert!((gini_impurity(&labels) - 0.0).abs() < 1e-7);
    }

    // ========================================================================
    // Gini Split Tests
    // ========================================================================

    #[test]
    fn test_gini_split_empty_both() {
        assert!((gini_split(&[], &[]) - 0.0).abs() < 1e-7);
    }

    #[test]
    fn test_gini_split_perfect_split() {
        // Perfect split: all 0s on left, all 1s on right -> weighted Gini = 0
        let left = vec![0, 0, 0];
        let right = vec![1, 1, 1];
        assert!((gini_split(&left, &right) - 0.0).abs() < 1e-7);
    }

    #[test]
    fn test_gini_split_worst_case() {
        // Both sides have same distribution -> split doesn't help
        let left = vec![0, 1];
        let right = vec![0, 1];
        assert!((gini_split(&left, &right) - 0.5).abs() < 1e-7);
    }

    // ========================================================================
    // Sorted Unique Values Tests
    // ========================================================================

    #[test]
    fn test_get_sorted_unique_values_basic() {
        let values = vec![3.0, 1.0, 2.0, 1.0, 3.0];
        let unique = get_sorted_unique_values(&values);
        assert_eq!(unique, vec![1.0, 2.0, 3.0]);
    }

    #[test]
    fn test_get_sorted_unique_values_single() {
        let values = vec![5.0];
        let unique = get_sorted_unique_values(&values);
        assert_eq!(unique, vec![5.0]);
    }

    #[test]
    fn test_get_sorted_unique_values_all_same() {
        let values = vec![2.0, 2.0, 2.0];
        let unique = get_sorted_unique_values(&values);
        assert_eq!(unique, vec![2.0]);
    }

    #[test]
    fn test_get_sorted_unique_values_already_sorted() {
        let values = vec![1.0, 2.0, 3.0, 4.0];
        let unique = get_sorted_unique_values(&values);
        assert_eq!(unique, vec![1.0, 2.0, 3.0, 4.0]);
    }

    // ========================================================================
    // Split Labels by Threshold Tests
    // ========================================================================

    #[test]
    fn test_split_labels_by_threshold_normal() {
        let x = vec![1.0, 2.0, 3.0, 4.0];
        let y = vec![0, 0, 1, 1];
        let result = split_labels_by_threshold(&x, &y, 2.5);
        let (left, right) = result.expect("split should produce non-empty partitions");
        assert_eq!(left, vec![0, 0]);
        assert_eq!(right, vec![1, 1]);
    }

    #[test]
    fn test_split_labels_by_threshold_all_left() {
        // All values <= threshold -> right is empty -> None
        let x = vec![1.0, 2.0, 3.0];
        let y = vec![0, 1, 0];
        assert!(split_labels_by_threshold(&x, &y, 10.0).is_none());
    }

    #[test]
    fn test_split_labels_by_threshold_all_right() {
        // All values > threshold -> left is empty -> None
        let x = vec![5.0, 6.0, 7.0];
        let y = vec![0, 1, 0];
        assert!(split_labels_by_threshold(&x, &y, 0.0).is_none());
    }

    #[test]
    fn test_split_labels_by_threshold_boundary() {
        // Exact threshold value goes to left (<=)
        let x = vec![1.0, 2.0, 3.0];
        let y = vec![0, 1, 2];
        let (left, right) = split_labels_by_threshold(&x, &y, 2.0).expect("split should succeed");
        assert_eq!(left, vec![0, 1]);
        assert_eq!(right, vec![2]);
    }

    // ========================================================================
    // Calculate Information Gain Tests
    // ========================================================================

    #[test]
    fn test_calculate_information_gain_perfect_split() {
        let current = gini_impurity(&[0, 0, 1, 1]);
        let gain = calculate_information_gain(current, &[0, 0], &[1, 1]);
        // Perfect split: gain equals the full impurity
        assert!((gain - current).abs() < 1e-7);
    }

    #[test]
    fn test_calculate_information_gain_no_improvement() {
        let current = gini_impurity(&[0, 1, 0, 1]);
        let gain = calculate_information_gain(current, &[0, 1], &[0, 1]);
        // Same distribution on both sides -> zero gain
        assert!(gain.abs() < 1e-7);
    }

    // ========================================================================
    // Find Best Split for Feature Tests
    // ========================================================================

    #[test]
    fn test_find_best_split_for_feature_too_few_samples() {
        let x = vec![1.0];
        let y = vec![0];
        assert!(find_best_split_for_feature(&x, &y).is_none());
    }

    #[test]
    fn test_find_best_split_for_feature_all_same_value() {
        let x = vec![1.0, 1.0, 1.0];
        let y = vec![0, 1, 0];
        assert!(find_best_split_for_feature(&x, &y).is_none());
    }

    #[test]
    fn test_find_best_split_for_feature_clear_split() {
        // Feature perfectly separates classes
        let x = vec![1.0, 2.0, 3.0, 4.0];
        let y = vec![0, 0, 1, 1];
        let result = find_best_split_for_feature(&x, &y);
        let (threshold, gain) = result.expect("should find a split");
        assert!((threshold - 2.5).abs() < 1e-7);
        assert!(gain > 0.0);
    }

    #[test]
    fn test_find_best_split_for_feature_pure_labels() {
        // All same label -> no gain possible
        let x = vec![1.0, 2.0, 3.0];
        let y = vec![0, 0, 0];
        assert!(find_best_split_for_feature(&x, &y).is_none());
    }

    // ========================================================================
    // Find Best Split (all features) Tests
    // ========================================================================

    #[test]
    fn test_find_best_split_single_sample() {
        let x = Matrix::from_vec(1, 2, vec![1.0, 2.0]).expect("matrix creation");
        let y = vec![0];
        assert!(find_best_split(&x, &y).is_none());
    }

    #[test]
    fn test_find_best_split_separable() {
        // Feature 0 separates perfectly, feature 1 does not
        let x = Matrix::from_vec(4, 2, vec![1.0, 5.0, 2.0, 5.0, 3.0, 5.0, 4.0, 5.0])
            .expect("matrix creation");
        let y = vec![0, 0, 1, 1];
        let (feat, threshold, gain) = find_best_split(&x, &y).expect("should find split");
        assert_eq!(feat, 0);
        assert!((threshold - 2.5).abs() < 1e-7);
        assert!(gain > 0.0);
    }

    #[test]
    fn test_find_best_split_pure_labels() {
        let x =
            Matrix::from_vec(3, 2, vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0]).expect("matrix creation");
        let y = vec![0, 0, 0];
        assert!(find_best_split(&x, &y).is_none());
    }

    // ========================================================================
    // Majority Class Tests
    // ========================================================================

    #[test]
    fn test_majority_class_single() {
        assert_eq!(majority_class(&[7]), 7);
    }

    #[test]
    fn test_majority_class_clear_winner() {
        assert_eq!(majority_class(&[0, 1, 1, 1, 0]), 1);
    }

    #[test]
    fn test_majority_class_all_same() {
        assert_eq!(majority_class(&[3, 3, 3]), 3);
    }

    // ========================================================================
    // Split Data by Indices Tests
    // ========================================================================

    #[test]
    fn test_split_data_by_indices_basic() {
        let x = Matrix::from_vec(4, 2, vec![1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0])
            .expect("matrix creation");
        let y = vec![0, 1, 2, 3];
        let indices = vec![0, 2];
        let (subset_x, subset_y) = split_data_by_indices(&x, &y, &indices);
        assert_eq!(subset_x.shape(), (2, 2));
        assert_eq!(subset_y, vec![0, 2]);
        assert!((subset_x.get(0, 0) - 1.0).abs() < 1e-7);
        assert!((subset_x.get(0, 1) - 2.0).abs() < 1e-7);
        assert!((subset_x.get(1, 0) - 5.0).abs() < 1e-7);
        assert!((subset_x.get(1, 1) - 6.0).abs() < 1e-7);
    }

    // ========================================================================
    // Check Stopping Criteria Tests
    // ========================================================================

    #[test]
    fn test_check_stopping_criteria_pure_node() {
        let y = vec![1, 1, 1];
        let result = check_stopping_criteria(&y, 0, None);
        let node = result.expect("pure node should trigger stop");
        match node {
            TreeNode::Leaf(leaf) => {
                assert_eq!(leaf.class_label, 1);
                assert_eq!(leaf.n_samples, 3);
            }
            TreeNode::Node(_) => panic!("expected leaf, got internal node"),
        }
    }

    #[test]
    fn test_check_stopping_criteria_max_depth() {
        let y = vec![0, 1, 1];
        let result = check_stopping_criteria(&y, 5, Some(5));
        let node = result.expect("max depth should trigger stop");
        match node {
            TreeNode::Leaf(leaf) => {
                assert_eq!(leaf.class_label, 1); // majority
                assert_eq!(leaf.n_samples, 3);
            }
            TreeNode::Node(_) => panic!("expected leaf, got internal node"),
        }
    }

    #[test]
    fn test_check_stopping_criteria_no_stop() {
        let y = vec![0, 1, 1];
        let result = check_stopping_criteria(&y, 2, Some(5));
        assert!(result.is_none());
    }

    #[test]
    fn test_check_stopping_criteria_no_max_depth() {
        let y = vec![0, 1, 1];
        let result = check_stopping_criteria(&y, 100, None);
        assert!(result.is_none());
    }

    // ========================================================================
    // Split Indices by Threshold Tests
    // ========================================================================

    #[test]
    fn test_split_indices_by_threshold_normal() {
        let x = Matrix::from_vec(4, 1, vec![1.0, 3.0, 2.0, 4.0]).expect("matrix creation");
        let result = split_indices_by_threshold(&x, 0, 2.5, 4);
        let (left, right) = result.expect("should split successfully");
        assert_eq!(left, vec![0, 2]);
        assert_eq!(right, vec![1, 3]);
    }

    #[test]
    fn test_split_indices_by_threshold_all_left() {
        let x = Matrix::from_vec(3, 1, vec![1.0, 2.0, 3.0]).expect("matrix creation");
        assert!(split_indices_by_threshold(&x, 0, 10.0, 3).is_none());
    }

    #[test]
    fn test_split_indices_by_threshold_all_right() {
        let x = Matrix::from_vec(3, 1, vec![5.0, 6.0, 7.0]).expect("matrix creation");
        assert!(split_indices_by_threshold(&x, 0, 0.0, 3).is_none());
    }

    // ========================================================================
    // Flatten / Reconstruct Tree Tests
    // ========================================================================

    #[test]
    fn test_flatten_and_reconstruct_leaf() {
        let leaf = TreeNode::Leaf(Leaf {
            class_label: 2,
            n_samples: 10,
        });

        let mut features = Vec::new();
        let mut thresholds = Vec::new();
        let mut classes = Vec::new();
        let mut samples = Vec::new();
        let mut left_children = Vec::new();
        let mut right_children = Vec::new();

        let root_idx = flatten_tree_node(
            &leaf,
            &mut features,
            &mut thresholds,
            &mut classes,
            &mut samples,
            &mut left_children,
            &mut right_children,
        );

        assert_eq!(root_idx, 0);
        assert_eq!(features.len(), 1);
        assert!((features[0] - (-1.0)).abs() < 1e-7);
        assert!((classes[0] - 2.0).abs() < 1e-7);
        assert!((samples[0] - 10.0).abs() < 1e-7);

        // Reconstruct
        let reconstructed = reconstruct_tree_node(
            root_idx,
            &features,
            &thresholds,
            &classes,
            &samples,
            &left_children,
            &right_children,
        );
        match reconstructed {
            TreeNode::Leaf(l) => {
                assert_eq!(l.class_label, 2);
                assert_eq!(l.n_samples, 10);
            }
            TreeNode::Node(_) => panic!("expected leaf"),
        }
    }