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
use std::marker::PhantomData;

use linfa::{Float, Label};
use ndarray::{Array1, ArrayView1, ArrayView2};
use serde::{ser::SerializeStruct, ser::Serializer, Deserialize, Deserializer, Serialize};

use crate::prelude::{Predictable, Trainable};

use super::Classifier;

/// A classifier that combines the predictions of two other classifiers.
///
/// # Type Parameters
///
/// * X - The type of the features in the training and prediction data.
/// * Y - The type of the labels in the training and prediction data.
/// * C1 - The type of the first classifier.
/// * C2 - The type of the second classifier.
#[derive(Clone, Debug, PartialEq)]
pub struct CombinedClassifier<X, Y, C1, C2>
where
    X: Float,
    Y: Label,
    C1: Classifier<X, Y>,
    C2: Classifier<X, Y>,
{
    x: PhantomData<*const X>,
    y: PhantomData<*const Y>,
    classifier1: C1,
    classifier2: C2,
}

impl<X, Y, C1, C2> CombinedClassifier<X, Y, C1, C2>
where
    X: Float,
    Y: Label,
    C1: Classifier<X, Y>,
    C2: Classifier<X, Y>,
{
    /// Creates a new CombinedClassifier
    ///
    /// # Arguments
    ///
    /// * classifier1 - The first classifier to use in the combination.
    /// * classifier2 - The second classifier to use in the combination.
    ///
    /// # Returns
    ///
    /// A new CombinedClassifier object.
    pub fn new(classifier1: C1, classifier2: C2) -> Self {
        Self {
            classifier1,
            classifier2,
            x: PhantomData,
            y: PhantomData,
        }
    }
}

impl<X, Y, C1, C2> Default for CombinedClassifier<X, Y, C1, C2>
where
    X: Float,
    Y: Label,
    C1: Classifier<X, Y> + Default,
    C2: Classifier<X, Y> + Default,
{
    fn default() -> Self {
        Self::new(C1::default(), C2::default())
    }
}

impl<X, Y, C1, C2> Trainable<X, Y> for CombinedClassifier<X, Y, C1, C2>
where
    X: Float,
    Y: Label,
    C1: Classifier<X, Y>,
    C2: Classifier<X, Y>,
{
    fn fit(&mut self, x: &ArrayView2<X>, y: &ArrayView1<Y>) -> Result<(), String> {
        // Train the first classifier
        self.classifier1.fit(x, y)?;

        // Train the second classifier
        self.classifier2.fit(x, y)?;

        Ok(())
    }
}

impl<X, Y, C1, C2> Predictable<X, Y> for CombinedClassifier<X, Y, C1, C2>
where
    X: Float,
    Y: Label + Copy,
    C1: Classifier<X, Y>,
    C2: Classifier<X, Y>,
{
    fn predict(&self, x: &ArrayView2<X>) -> Result<Array1<Y>, String> {
        // Make predictions using the first classifier
        let predictions1 = self.classifier1.predict(x)?;

        // Make predictions using the second classifier
        let predictions2 = self.classifier2.predict(x)?;

        // Combine the predictions by taking the intersection
        let combined_predictions = predictions1
            .iter()
            .zip(predictions2.iter())
            .map(|(p1, p2)| if p1 == p2 { *p1 } else { Y::default() })
            .collect();

        Ok(combined_predictions)
    }
}

impl<X, Y, C1, C2> Classifier<X, Y> for CombinedClassifier<X, Y, C1, C2>
where
    X: Float,
    Y: Label + Copy,
    C1: Classifier<X, Y>,
    C2: Classifier<X, Y>,
{
}

impl<X, Y, C1, C2> Serialize for CombinedClassifier<X, Y, C1, C2>
where
    X: Float,
    Y: Label,
    C1: Classifier<X, Y> + Serialize,
    C2: Classifier<X, Y> + Serialize,
{
    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
    where
        S: Serializer,
    {
        let mut state = serializer.serialize_struct("CombinedClassifier", 2)?;
        state.serialize_field("classifier1", &self.classifier1)?;
        state.serialize_field("classifier2", &self.classifier2)?;
        state.end()
    }
}

impl<'de, X, Y, C1, C2> Deserialize<'de> for CombinedClassifier<X, Y, C1, C2>
where
    X: Float,
    Y: Label,
    C1: Classifier<X, Y> + Deserialize<'de>,
    C2: Classifier<X, Y> + Deserialize<'de>,
{
    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
    where
        D: Deserializer<'de>,
    {
        #[derive(Deserialize)]
        struct CombinedClassifierData<C1, C2> {
            classifier1: C1,
            classifier2: C2,
        }

        let combined_classifier_data = CombinedClassifierData::deserialize(deserializer)?;
        Ok(CombinedClassifier {
            x: PhantomData,
            y: PhantomData,
            classifier1: combined_classifier_data.classifier1,
            classifier2: combined_classifier_data.classifier2,
        })
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::prelude::{BayesClassifier, RandomForestClassifier, SVMClassifier};
    use ndarray::array;

    #[test]
    fn test_combined_classifier() {
        // Create an instance of CombinedClassifier using two classifier implementations
        let mut combined_classifier: CombinedClassifier<
            f32,
            bool,
            SVMClassifier<_, _>,
            BayesClassifier<_, _>,
        > = CombinedClassifier::default();

        // Generate training data and labels
        let x_train = array![[1., 1.], [0., 2.], [1., 3.], [1., 4.], [0., 5.], [1., 6.]];
        let y_train = array![true, false, true, true, false, true];

        // Fit the CombinedClassifier using the training data and labels
        combined_classifier
            .fit(&x_train.view(), &y_train.view())
            .unwrap();

        // Generate test data
        let x_test = array![[1., 1.], [0., 2.]];

        // Make predictions using the CombinedClassifier
        let y_pred = combined_classifier.predict(&x_test.view()).unwrap();

        // Assert that the predictions are correct
        assert_eq!(y_pred, array![true, false]);
    }

    #[test]
    fn test_combined_classifier_serialization() {
        // Create an instance of CombinedClassifier using two classifier implementations
        let combined_classifier: CombinedClassifier<
            f32,
            usize,
            RandomForestClassifier<_, _>,
            RandomForestClassifier<_, _>,
        > = CombinedClassifier::default();

        let serialized = serde_json::to_string(&combined_classifier).unwrap();

        let deserialized: CombinedClassifier<
            f32,
            usize,
            RandomForestClassifier<f32, usize>,
            RandomForestClassifier<f32, usize>,
        > = serde_json::from_str(&serialized).unwrap();

        assert_eq!(combined_classifier, deserialized);
    }
}