perpetual 2.0.0

A self-generalizing gradient boosting machine that doesn't need hyperparameter optimization
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
247
248
249
250
251
252
253

//! Huber Loss function for robust regression.

use crate::{objective::ObjectiveFunction, utils::weighted_median};
use serde::{Deserialize, Serialize};

/// Huber Loss — quadratic near zero, linear beyond `delta`.
#[derive(Default, Debug, Deserialize, Serialize, Clone)]
pub struct HuberLoss {
    pub delta: Option<f64>,
}

impl ObjectiveFunction for HuberLoss {
    #[inline]
    fn loss(&self, y: &[f64], yhat: &[f64], sample_weight: Option<&[f64]>, _group: Option<&[u64]>) -> Vec<f32> {
        let delta = self.delta.unwrap_or(1.0);
        match sample_weight {
            Some(weights) => y
                .iter()
                .zip(yhat.iter())
                .enumerate()
                .map(|(i, (&yi, &yh))| {
                    let r = yi - yh;
                    let ar = r.abs();
                    let base = if ar <= delta {
                        0.5 * r * r
                    } else {
                        delta * (ar - 0.5 * delta)
                    };
                    (base * weights[i]) as f32
                })
                .collect(),
            None => y
                .iter()
                .zip(yhat.iter())
                .map(|(&yi, &yh)| {
                    let r = yi - yh;
                    let ar = r.abs();
                    let loss = if ar <= delta {
                        0.5 * r * r
                    } else {
                        delta * (ar - 0.5 * delta)
                    };
                    loss as f32
                })
                .collect(),
        }
    }

    #[inline]
    fn gradient(
        &self,
        y: &[f64],
        yhat: &[f64],
        sample_weight: Option<&[f64]>,
        _group: Option<&[u64]>,
    ) -> (Vec<f32>, Option<Vec<f32>>) {
        let delta = self.delta.unwrap_or(1.0) as f32;
        let len = y.len();
        let mut g = Vec::with_capacity(len);
        let mut h = Vec::with_capacity(len);

        match sample_weight {
            Some(weights) => {
                for i in 0..len {
                    let diff = (yhat[i] - y[i]) as f32;
                    let w = weights[i] as f32;
                    if diff.abs() <= delta {
                        g.push(diff * w);
                        h.push(w);
                    } else {
                        g.push(delta * diff.signum() * w);
                        h.push(0.0);
                    }
                }
                (g, Some(h))
            }
            None => {
                for i in 0..len {
                    let diff = (yhat[i] - y[i]) as f32;
                    if diff.abs() <= delta {
                        g.push(diff);
                        h.push(1.0);
                    } else {
                        g.push(delta * diff.signum());
                        h.push(0.0);
                    }
                }
                (g, Some(h))
            }
        }
    }

    #[inline]
    fn initial_value(&self, y: &[f64], sample_weight: Option<&[f64]>, _group: Option<&[u64]>) -> f64 {
        weighted_median(y, sample_weight)
    }

    // `default_metric`: inherits the trait default (`RootMeanSquaredError`).

    fn gradient_and_loss(
        &self,
        y: &[f64],
        yhat: &[f64],
        sample_weight: Option<&[f64]>,
        _group: Option<&[u64]>,
    ) -> (Vec<f32>, Option<Vec<f32>>, Vec<f32>) {
        let delta = self.delta.unwrap_or(1.0);
        let delta32 = delta as f32;
        let len = y.len();
        let mut g = Vec::with_capacity(len);
        let mut h = Vec::with_capacity(len);
        let mut l = Vec::with_capacity(len);

        match sample_weight {
            Some(weights) => {
                for i in 0..len {
                    let r = yhat[i] - y[i];
                    let ar = r.abs();
                    let diff = r as f32;
                    let w = weights[i] as f32;
                    if ar <= delta {
                        g.push(diff * w);
                        h.push(w);
                        l.push((0.5 * r * r * weights[i]) as f32);
                    } else {
                        g.push(delta32 * diff.signum() * w);
                        h.push(0.0);
                        l.push((delta * (ar - 0.5 * delta) * weights[i]) as f32);
                    }
                }
            }
            None => {
                for i in 0..len {
                    let r = yhat[i] - y[i];
                    let ar = r.abs();
                    let diff = r as f32;
                    if ar <= delta {
                        g.push(diff);
                        h.push(1.0);
                        l.push((0.5 * r * r) as f32);
                    } else {
                        g.push(delta32 * diff.signum());
                        h.push(0.0);
                        l.push((delta * (ar - 0.5 * delta)) as f32);
                    }
                }
            }
        }
        (g, Some(h), l)
    }

    fn requires_batch_evaluation(&self) -> bool {
        false
    }
}

impl HuberLoss {
    #[inline]
    pub fn loss_single(&self, y: f64, yhat: f64, sample_weight: Option<f64>) -> f32 {
        let delta = self.delta.unwrap_or(1.0);
        let r = y - yhat;
        let ar = r.abs();
        let l = if ar <= delta {
            0.5 * r * r
        } else {
            delta * (ar - 0.5 * delta)
        };
        match sample_weight {
            Some(w) => (l * w) as f32,
            None => l as f32,
        }
    }
}
#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_huber_loss() {
        let y = vec![1.0, 2.0];
        let yhat = vec![1.1, 3.0]; // diffs: 0.1, 1.0. Delta default is 1.0.
        let loss_fn = HuberLoss { delta: Some(1.0) };

        // Test loss
        let l = loss_fn.loss(&y, &yhat, None, None);
        assert_eq!(l.len(), 2);
        assert!((l[0] - 0.5 * 0.1 * 0.1).abs() < 1e-6); // quadratic
        assert!((l[1] - 0.5).abs() < 1e-6); // linear boundary (1.0 * (1.0 - 0.5))

        // Test gradient/hessian
        let (g, h) = loss_fn.gradient(&y, &yhat, None, None);
        let h = h.unwrap();
        // g = yhat - y for |diff| <= delta
        assert!((g[0] - 0.1).abs() < 1e-6);
        assert!((h[0] - 1.0).abs() < 1e-6);
        // g = delta * sign(yhat - y) for |diff| > delta
        // at diff=1.0, sign is 1.0, g=1.0*1.0=1.0
        assert!((g[1] - 1.0).abs() < 1e-6);
        assert!((h[1] - 1.0).abs() < 1e-6);
    }

    #[test]
    fn test_huber_loss_weighted() {
        let y = vec![1.0];
        let yhat = vec![2.0]; // diff = 1.0
        let weights = vec![0.5];
        let loss_fn = HuberLoss { delta: Some(0.5) }; // delta < diff, so linear

        let l = loss_fn.loss(&y, &yhat, Some(&weights), None);
        // loss = delta * (diff - 0.5*delta) * weight = 0.5 * (1.0 - 0.25) * 0.5 = 0.5 * 0.75 * 0.5 = 0.1875
        assert!((l[0] - 0.1875).abs() < 1e-6);

        let (g, h) = loss_fn.gradient(&y, &yhat, Some(&weights), None);
        let h = h.unwrap();
        // g = delta * sign(diff) * weight = 0.5 * 1.0 * 0.5 = 0.25
        assert!((g[0] - 0.25).abs() < 1e-6);
        assert!((h[0] - 0.0).abs() < 1e-6);
    }

    #[test]
    fn test_huber_gradient_and_loss() {
        let y = vec![1.0, 2.0];
        let yhat = vec![1.1, 3.0];
        let loss_fn = HuberLoss { delta: Some(1.0) };
        let (g, h, l) = loss_fn.gradient_and_loss(&y, &yhat, None, None);
        assert_eq!(g.len(), 2);
        assert!(h.is_some());
        assert_eq!(l.len(), 2);
    }

    #[test]
    fn test_huber_gradient_and_loss_weighted() {
        let y = vec![1.0, 2.0];
        let yhat = vec![1.1, 3.0];
        let w = vec![2.0, 1.0];
        let loss_fn = HuberLoss { delta: Some(1.0) };
        let (g, h, l) = loss_fn.gradient_and_loss(&y, &yhat, Some(&w), None);
        assert_eq!(g.len(), 2);
        assert!(h.is_some());
        assert_eq!(l.len(), 2);
    }

    #[test]
    fn test_huber_loss_single() {
        let loss_fn = HuberLoss { delta: Some(1.0) };
        let l1 = loss_fn.loss_single(1.0, 1.5, None); // r=0.5, ar=0.5<=1 -> 0.5*0.25=0.125
        assert!((l1 - 0.125).abs() < 1e-5);
        let l2 = loss_fn.loss_single(1.0, 3.0, None); // r=2, ar=2>1 -> 1*(2-0.5)=1.5
        assert!((l2 - 1.5).abs() < 1e-5);
        let l3 = loss_fn.loss_single(1.0, 1.5, Some(2.0));
        assert!((l3 - 0.25).abs() < 1e-5);
    }
}