random forest

random_forest.py

from sklearn.base import ClassifierMixin
from collections import Counter
import numpy as np

class RandomForest():
    def __init__(self, n_trees=10):
        self._n_trees = n_trees

        self._forest = [None] * self._n_trees
        self._using_data = [None] * self._n_trees

    def _bootstrap_sample(self, X, y):
        n_data = X.shape[1]
        n_data_forest = np.floor(np.sqrt(n_data))
        bootstrapped_X = list()
        bootstrapped_y = list()
        for i in range(self._n_trees):
            index = np.random.choice(len(y), size=len(y))
            col = np.random.choice(n_data, size=n_data_forest, replace=False)
            bootstrapped_X.append(X[np.ix_(index, col)])
            bootstrapped_y.append(y[index])
            self._using_data[i] = col
        return bootstrapped_X, bootstrapped_y

    def create_forest(self, X, y):
        self._targets = np.unique(y)

        bootstrap_X, bootstrapped_y = self._bootstrap_sample(X, y)
        for i, (i_bootstrapped_X, i_bootstrapped_y) in enumerate(zip(bootstrapped_X, bootstrapped_y)):
            tree = DecisionTree()
            tree.create_tree(i_bootstrapped_X, i_bootstrapped_y)
            self._forest[i] = tree

    def predict(self, X):
        proba = self._predict_proba(X)
        return self._targets[np.argmax(proba, axis=1)]

    def _predict_proba(self, X):
        if self._forest[0] is None:
            raise ValueError('fitしてね')
        votes = [tree.predict(X[:, using_data]) for tree, using_data in zip(self._forest, self._using_features)]
        counts = [Counter(row) for row in np.array(votes).transpose()]