fix: minor refactoring

2020-03-20 15:58:10 -07:00
parent a96f303dea
commit 6577e22111
3 changed files with 19 additions and 19 deletions
@@ -0,0 +1,473 @@
+use std::default::Default;
+use std::collections::LinkedList;
+use crate::linalg::Matrix;
+use crate::algorithm::sort::quick_sort::QuickArgSort;
+
+#[derive(Debug)]
+pub struct DecisionTreeClassifierParameters {           
+    pub criterion: SplitCriterion,
+    pub max_depth: Option<u16>,
+    pub min_samples_leaf: u16    
+}
+
+#[derive(Debug)]
+pub struct DecisionTreeClassifier {    
+    nodes: Vec<Node>,    
+    parameters: DecisionTreeClassifierParameters,    
+    num_classes: usize,
+    classes: Vec<f64>,
+    depth: u16
+}
+
+#[derive(Debug, Clone)]
+pub enum SplitCriterion {
+    Gini,
+    Entropy,
+    ClassificationError
+}
+
+#[derive(Debug)]
+pub struct Node {
+    index:  usize,
+    output: usize,    
+    split_feature: usize,
+    split_value: f64,
+    split_score: f64,
+    true_child: Option<usize>,
+    false_child: Option<usize>,    
+}
+
+
+impl Default for DecisionTreeClassifierParameters {
+    fn default() -> Self { 
+        DecisionTreeClassifierParameters {
+            criterion: SplitCriterion::Gini,
+            max_depth: None,
+            min_samples_leaf: 1
+        }
+     }
+}
+
+impl Node {
+    fn new(index: usize, output: usize) -> Self { 
+        Node {
+            index:  index,
+            output: output,
+            split_feature: 0,
+            split_value: std::f64::NAN,
+            split_score: std::f64::NAN,
+            true_child: Option::None,
+            false_child: Option::None            
+        }
+     }
+}
+
+struct NodeVisitor<'a, M: Matrix> {
+    x: &'a M,
+    y: &'a Vec<usize>,
+    node: usize,
+    samples: Vec<u32>,
+    order: &'a Vec<Vec<usize>>, 
+    true_child_output: usize,
+    false_child_output: usize,
+    level: u16
+}
+
+fn impurity(criterion: &SplitCriterion, count: &Vec<u32>, n: u32) -> f64 {
+    let mut impurity = 0.;
+
+    match criterion {
+        SplitCriterion::Gini => {
+            impurity = 1.0;
+            for i in 0..count.len() {
+                if count[i] > 0 {
+                    let p = count[i] as f64 / n as f64;
+                    impurity -= p * p;
+                }
+            }                
+        }
+
+        SplitCriterion::Entropy => {
+            for i in 0..count.len() {
+                if count[i] > 0 {
+                    let p = count[i] as f64 / n as f64;
+                    impurity -= p * p.log2();
+                }
+            }
+        }
+        SplitCriterion::ClassificationError => {
+            for i in 0..count.len() {
+                if count[i] > 0 {
+                    impurity = impurity.max(count[i] as f64 / n as f64);
+                }
+            }
+            impurity = (1. - impurity).abs();
+        }                
+    }
+
+    return impurity;
+}
+
+impl<'a, M: Matrix> NodeVisitor<'a, M> {    
+
+    fn new(node_id: usize, samples: Vec<u32>, order: &'a Vec<Vec<usize>>, x: &'a M, y: &'a Vec<usize>, level: u16) -> Self {
+        NodeVisitor {
+            x: x,
+            y: y,
+            node: node_id,
+            samples: samples,
+            order: order,
+            true_child_output: 0,
+            false_child_output: 0,
+            level: level
+        }
+    }
+
+}
+
+pub(in crate) fn which_max(x: &Vec<u32>) -> usize {
+    let mut m = x[0];
+    let mut which = 0;
+
+    for i in 1..x.len() {
+        if x[i] > m {
+            m = x[i];
+            which = i;
+        }
+    }
+
+    return which;
+}
+
+impl DecisionTreeClassifier {
+
+    pub fn fit<M: Matrix>(x: &M, y: &M::RowVector, parameters: DecisionTreeClassifierParameters) -> DecisionTreeClassifier {
+        let (x_nrows, num_attributes) = x.shape();
+        let samples = vec![1; x_nrows];
+        DecisionTreeClassifier::fit_weak_learner(x, y, samples, num_attributes, parameters)
+    }
+
+    pub fn fit_weak_learner<M: Matrix>(x: &M, y: &M::RowVector, samples: Vec<u32>, mtry: usize, parameters: DecisionTreeClassifierParameters) -> DecisionTreeClassifier {
+        let y_m = M::from_row_vector(y.clone());
+        let (_, y_ncols) = y_m.shape();
+        let (_, num_attributes) = x.shape();
+        let classes = y_m.unique();        
+        let k = classes.len(); 
+        if k < 2 {
+            panic!("Incorrect number of classes: {}. Should be >= 2.", k);
+        }
+
+        let mut yi: Vec<usize> = vec![0; y_ncols];
+
+        for i in 0..y_ncols {
+            let yc = y_m.get(0, i);                        
+            yi[i] = classes.iter().position(|c| yc == *c).unwrap();
+        }
+
+        let mut nodes: Vec<Node> = Vec::new();        
+
+        let mut count = vec![0; k];
+        for i in 0..y_ncols {
+            count[yi[i]] += samples[i];
+        }        
+
+        let root = Node::new(0, which_max(&count));        
+        nodes.push(root);
+        let mut order: Vec<Vec<usize>> = Vec::new();
+
+        for i in 0..num_attributes {
+            order.push(x.get_col_as_vec(i).quick_argsort());
+        }                        
+
+        let mut tree = DecisionTreeClassifier{                                       
+            nodes: nodes,            
+            parameters: parameters,    
+            num_classes: k,
+            classes: classes,
+            depth: 0        
+        };
+
+        let mut visitor = NodeVisitor::<M>::new(0, samples, &order, &x, &yi, 1);
+
+        let mut visitor_queue: LinkedList<NodeVisitor<M>> = LinkedList::new();
+
+        if tree.find_best_cutoff(&mut visitor, mtry) {
+            visitor_queue.push_back(visitor);
+        }
+
+        while tree.depth < tree.parameters.max_depth.unwrap_or(std::u16::MAX) {            
+            match visitor_queue.pop_front() {
+                Some(node) => tree.split(node, mtry, &mut visitor_queue,),
+                None => break
+            };     
+        }        
+
+        tree
+    }
+
+    pub fn predict<M: Matrix>(&self, x: &M) -> M::RowVector {
+        let mut result = M::zeros(1, x.shape().0);
+
+        let (n, _) = x.shape();
+
+        for i in 0..n {
+            result.set(0, i, self.classes[self.predict_for_row(x, i)]);
+        }
+
+        result.to_row_vector()
+    }
+
+    pub(in crate) fn predict_for_row<M: Matrix>(&self, x: &M, row: usize) -> usize {
+        let mut result = 0;
+        let mut queue: LinkedList<usize> = LinkedList::new();
+
+        queue.push_back(0);
+        
+        while !queue.is_empty() {
+            match queue.pop_front() {
+                Some(node_id) => {
+                    let node = &self.nodes[node_id];
+                    if node.true_child == None && node.false_child == None {
+                        result = node.output;
+                    } else {
+                        if x.get(row, node.split_feature) <= node.split_value {
+                            queue.push_back(node.true_child.unwrap());
+                        } else {
+                            queue.push_back(node.false_child.unwrap());
+                        }
+                    }
+                },
+                None => break
+            };
+        }
+
+        return result
+        
+    }   
+    
+    fn find_best_cutoff<M: Matrix>(&mut self, visitor: &mut NodeVisitor<M>, mtry: usize) -> bool {
+
+        let (n_rows, n_attr) = visitor.x.shape();        
+
+        let mut label = Option::None;
+        let mut is_pure = true;
+        for i in 0..n_rows {
+            if visitor.samples[i] > 0 {
+                if label == Option::None {
+                    label = Option::Some(visitor.y[i]);
+                } else if visitor.y[i] != label.unwrap() {
+                    is_pure = false;
+                    break;
+                }
+            }
+        }
+                
+        if is_pure {
+            return false;
+        }
+
+        let n = visitor.samples.iter().sum();        
+
+        if n <= self.parameters.min_samples_leaf as u32 {
+            return false;
+        }
+        
+        let mut count = vec![0; self.num_classes];
+        let mut false_count = vec![0; self.num_classes];
+        for i in 0..n_rows {
+            if visitor.samples[i] > 0 {
+                count[visitor.y[i]] += visitor.samples[i];
+            }
+        }
+
+        let parent_impurity = impurity(&self.parameters.criterion, &count, n);
+                
+        let mut variables = vec![0; n_attr];
+        for i in 0..n_attr {
+            variables[i] = i;
+        }
+
+        for j in 0..mtry {
+            self.find_best_split(visitor, n, &count, &mut false_count, parent_impurity, variables[j]);            
+        }        
+
+        !self.nodes[visitor.node].split_score.is_nan()
+
+    }    
+
+    fn find_best_split<M: Matrix>(&mut self, visitor: &mut NodeVisitor<M>, n: u32, count: &Vec<u32>, false_count: &mut Vec<u32>, parent_impurity: f64, j: usize){
+
+        let mut true_count = vec![0; self.num_classes];
+        let mut prevx = std::f64::NAN;
+        let mut prevy = 0;
+        let node_size = 1;
+
+        for i in visitor.order[j].iter() {
+            if visitor.samples[*i] > 0 {
+                if prevx.is_nan() || visitor.x.get(*i, j) == prevx || visitor.y[*i] == prevy {
+                    prevx = visitor.x.get(*i, j);
+                    prevy = visitor.y[*i];
+                    true_count[visitor.y[*i]] += visitor.samples[*i];
+                    continue;
+                }
+
+                let tc = true_count.iter().sum();
+                let fc = n - tc;
+                
+                if tc < node_size || fc < node_size {
+                    prevx = visitor.x.get(*i, j);
+                    prevy = visitor.y[*i];
+                    true_count[visitor.y[*i]] += visitor.samples[*i];
+                    continue;
+                }
+
+                for l in 0..self.num_classes {
+                    false_count[l] = count[l] - true_count[l];
+                }
+
+                let true_label = which_max(&true_count);
+                let false_label = which_max(false_count);                
+                let gain = parent_impurity - tc as f64 / n as f64 * impurity(&self.parameters.criterion, &true_count, tc) - fc as f64 / n as f64 * impurity(&self.parameters.criterion, &false_count, fc);
+
+                if self.nodes[visitor.node].split_score.is_nan() || gain > self.nodes[visitor.node].split_score {                    
+                    self.nodes[visitor.node].split_feature = j;
+                    self.nodes[visitor.node].split_value = (visitor.x.get(*i, j) + prevx) / 2.;
+                    self.nodes[visitor.node].split_score = gain;
+                    visitor.true_child_output = true_label;
+                    visitor.false_child_output = false_label;
+                }
+
+                prevx = visitor.x.get(*i, j);
+                prevy = visitor.y[*i];
+                true_count[visitor.y[*i]] += visitor.samples[*i];
+            }
+        }
+
+    }
+
+    fn split<'a, M: Matrix>(&mut self, mut visitor: NodeVisitor<'a, M>, mtry: usize, visitor_queue: &mut LinkedList<NodeVisitor<'a, M>>) -> bool {
+        let (n, _) = visitor.x.shape();
+        let mut tc = 0;
+        let mut fc = 0;        
+        let mut true_samples: Vec<u32> = vec![0; n];
+
+        for i in 0..n {
+            if visitor.samples[i] > 0 {
+                if visitor.x.get(i, self.nodes[visitor.node].split_feature) <= self.nodes[visitor.node].split_value {
+                    true_samples[i] = visitor.samples[i];
+                    tc += true_samples[i];
+                    visitor.samples[i] = 0;
+                } else {                    
+                    fc += visitor.samples[i];
+                }
+            }
+        }
+
+        if tc < self.parameters.min_samples_leaf as u32 || fc < self.parameters.min_samples_leaf as u32 {
+            self.nodes[visitor.node].split_feature = 0;
+            self.nodes[visitor.node].split_value = std::f64::NAN;
+            self.nodes[visitor.node].split_score = std::f64::NAN;
+            return false;
+        }        
+
+        let true_child_idx = self.nodes.len();
+        self.nodes.push(Node::new(true_child_idx, visitor.true_child_output));
+        let false_child_idx = self.nodes.len();
+        self.nodes.push(Node::new(false_child_idx, visitor.false_child_output));
+
+        self.nodes[visitor.node].true_child = Some(true_child_idx);
+        self.nodes[visitor.node].false_child = Some(false_child_idx);
+        
+        self.depth = u16::max(self.depth, visitor.level + 1);
+
+        let mut true_visitor = NodeVisitor::<M>::new(true_child_idx, true_samples, visitor.order, visitor.x, visitor.y, visitor.level + 1);            
+            
+        if tc > self.parameters.min_samples_leaf as u32 && self.find_best_cutoff(&mut true_visitor, mtry) {
+            visitor_queue.push_back(true_visitor);
+        }
+
+        let mut false_visitor = NodeVisitor::<M>::new(false_child_idx, visitor.samples, visitor.order, visitor.x, visitor.y, visitor.level + 1);
+            
+        if fc > self.parameters.min_samples_leaf as u32 && self.find_best_cutoff(&mut false_visitor, mtry) {
+            visitor_queue.push_back(false_visitor);
+        }
+
+        true
+    }
+
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*; 
+    use crate::linalg::naive::dense_matrix::DenseMatrix;
+
+    #[test]
+    fn gini_impurity() {
+        assert!((impurity(&SplitCriterion::Gini, &vec![7, 3], 10) - 0.42).abs() < std::f64::EPSILON);
+        assert!((impurity(&SplitCriterion::Entropy, &vec![7, 3], 10) - 0.8812908992306927).abs() < std::f64::EPSILON);
+        assert!((impurity(&SplitCriterion::ClassificationError, &vec![7, 3], 10) - 0.3).abs() < std::f64::EPSILON);
+    }
+
+    #[test]
+    fn fit_predict_iris() {             
+
+        let x = DenseMatrix::from_array(&[
+            &[5.1, 3.5, 1.4, 0.2],
+            &[4.9, 3.0, 1.4, 0.2],
+            &[4.7, 3.2, 1.3, 0.2],
+            &[4.6, 3.1, 1.5, 0.2],
+            &[5.0, 3.6, 1.4, 0.2],
+            &[5.4, 3.9, 1.7, 0.4],
+            &[4.6, 3.4, 1.4, 0.3],
+            &[5.0, 3.4, 1.5, 0.2],
+            &[4.4, 2.9, 1.4, 0.2],
+            &[4.9, 3.1, 1.5, 0.1],
+            &[7.0, 3.2, 4.7, 1.4],
+            &[6.4, 3.2, 4.5, 1.5],
+            &[6.9, 3.1, 4.9, 1.5],
+            &[5.5, 2.3, 4.0, 1.3],
+            &[6.5, 2.8, 4.6, 1.5],
+            &[5.7, 2.8, 4.5, 1.3],
+            &[6.3, 3.3, 4.7, 1.6],
+            &[4.9, 2.4, 3.3, 1.0],
+            &[6.6, 2.9, 4.6, 1.3],
+            &[5.2, 2.7, 3.9, 1.4]]);
+        let y = vec![0., 0., 0., 0., 0., 0., 0., 0., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1., 1.];
+
+        assert_eq!(y, DecisionTreeClassifier::fit(&x, &y, Default::default()).predict(&x));
+        
+        assert_eq!(3, DecisionTreeClassifier::fit(&x, &y, DecisionTreeClassifierParameters{criterion: SplitCriterion::Entropy, max_depth: Some(3), min_samples_leaf: 1}).depth);        
+            
+    }
+
+    #[test]
+    fn fit_predict_baloons() {             
+
+        let x = DenseMatrix::from_array(&[
+            &[1.,1.,1.,0.],
+            &[1.,1.,1.,0.],
+            &[1.,1.,1.,1.],
+            &[1.,1.,0.,0.],
+            &[1.,1.,0.,1.],
+            &[1.,0.,1.,0.],
+            &[1.,0.,1.,0.],
+            &[1.,0.,1.,1.],
+            &[1.,0.,0.,0.],
+            &[1.,0.,0.,1.],
+            &[0.,1.,1.,0.],
+            &[0.,1.,1.,0.],
+            &[0.,1.,1.,1.],
+            &[0.,1.,0.,0.],
+            &[0.,1.,0.,1.],
+            &[0.,0.,1.,0.],
+            &[0.,0.,1.,0.],
+            &[0.,0.,1.,1.],
+            &[0.,0.,0.,0.],
+            &[0.,0.,0.,1.]]);
+        let y = vec![1., 1., 0., 0., 0., 1., 1., 0., 0., 0., 1., 1., 0., 0., 0., 1., 1., 0., 0., 0.];
+
+        assert_eq!(y, DecisionTreeClassifier::fit(&x, &y, Default::default()).predict(&x));
+            
+    }
+}