数据挖掘——月亮数据

一、问题描述

月亮数据是sklearn工具库提供的一个数据集。它上用于分类和聚类算法的实践实验。图中每一个点是一条数据。其中(x1,x2)是特征组，颜色是标签值。    

二、实验目的

学习决策树和随机森林

三、实验内容

1.数据导入：采用自动生成的数据

2.数据预处理：使用库函数进行数据处理

四、实验结果及分析

原始数据：

Max_depth=2：

Max_depth = 5：

五、遇到的问题和解决方法

图像处理的时候不太懂，参考别人的做的。

六、完整代码

decisionTreeBase.py

import numpy as np
from machine_learning.homework.week10.TreeNode import Nodeclass DecisionTreeBase:def __init__(self, max_depth, feature_sample_rate, get_score):self.max_depth = max_depthself.feature_sample_rate = feature_sample_rateself.get_score = get_scoredef split_data(self, j, theta, X, idx):idx1, idx2 = list(), list()for i in idx:value = X[i][j]if value <= theta:idx1.append(i)else:idx2.append(i)return idx1, idx2def get_random_features(self, n):shuffled = np.random.permutation(n)size = int(self.feature_sample_rate * n)selected = shuffled[:size]return selecteddef find_best_split(self, X, y, idx):m, n = X.shapebest_score = float("inf")best_j = -1best_theta = float("inf")best_idx1, best_idx2 = list(), list()selected_j = self.get_random_features(n)for j in selected_j:thetas = set([x[j] for x in X])for theta in thetas:idx1, idx2 = self.split_data(j, theta, X, idx)if min(len(idx1), len(idx2)) == 0:continuescore1, score2 = self.get_score(y, idx1), self.get_score(y, idx2)w = 1.0 * len(idx1) / len(idx)score = w * score1 + (1 - w) * score2if score < best_score:best_score = scorebest_j = jbest_theta = thetabest_idx1 = idx1best_idx2 = idx2return best_j, best_theta, best_idx1, best_idx2, best_scoredef generate_tree(self, X, y, idx, d):r = Node()r.p = np.average(y[idx], axis=0)if d == 0 or len(idx) < 2:return rcurrent_score = self.get_score(y, idx)j, theta, idx1, idx2, score = self.find_best_split(X, y, idx)if score >= current_score:return rr.j = jr.theta = thetar.left = self.generate_tree(X, y, idx1, d - 1)r.right = self.generate_tree(X, y, idx2, d - 1)return rdef fit(self, X, y):self.root = self.generate_tree(X, y, range(len(X)), self.max_depth)def get_prediction(self, r, x):if r.left == None and r.right == None:return r.pvalue = x[r.j]if value <= r.theta:return self.get_prediction(r.left, x)else:return self.get_prediction(r.right, x)def predict(self, X):y = list()for i in range(len(X)):y.append(self.get_prediction(self.root, X[i]))return np.array(y)

decisionTreeClassifier.py

import numpy as np
from machine_learning.homework.week10.decisionTreeBase import DecisionTreeBasedef get_impurity(y, idx):p = np.average(y[idx], axis=0)return 1 - p.dot(p.T)def get_entropy(y, idx):_, k = y.shapep = np.average(y[idx], axis=0)return - np.log(p + 0.001 * np.random.rand(k)).dot(p.T)class DecisionTreeClassifier(DecisionTreeBase):def __init__(self, max_depth=0, feature_sample_rate=1.0):super().__init__(max_depth=max_depth,feature_sample_rate=feature_sample_rate,get_score=get_entropy)def predict_proba(self, X):return super().predict(X)def predict(self, X):proba = self.predict_proba(X)return np.argmax(proba, axis=1)

moon.py

from sklearn.datasets import make_moons
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from machine_learning.homework.week10.decisionTreeClassifier import DecisionTreeClassifier
from machine_learning.homework.week10.randomForestClassifier import RandomForestClassifier
from sklearn.metrics import accuracy_score
import numpy as npdef convert_to_vector(y):m = len(y)k = np.max(y) + 1v = np.zeros(m * k).reshape(m,k)for i in range(m):v[i][y[i]] = 1return vX, y = make_moons(n_samples=1000, noise=0.1)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.5, random_state=0)
plt.figure(0)
plt.axis([-1.5, 2.5, -0.75, 1.25 ])
plt.scatter(X_train[:, 0][y_train==0], X_train[:, 1][y_train==0], c='b', marker='o', s=10)
plt.scatter(X_train[:, 0][y_train==1], X_train[:, 1][y_train==1], c='r', marker='o', s=10)plt.show()tree = DecisionTreeClassifier(max_depth=5)
tree.fit(X_train, convert_to_vector(y_train))
y_pred = tree.predict(X_test)
print("decision tree accuracy= {}".format(accuracy_score(y_test, y_pred)))plt.figure(1)
x0s = np.linspace(-3, 4, 100)
x1s = np.linspace(-1, 6, 100)
x0, x1 = np.meshgrid(x0s, x1s)
W = np.c_[x0.ravel(), x1.ravel()]
u= tree.predict(W).reshape(x0.shape)
plt.axis([-1.5, 2.5, -0.75, 1.25 ])
plt.scatter(X_train[:, 0][y_train==0], X_train[:, 1][y_train==0], c='b', marker='o', s=30)
plt.scatter(X_train[:, 0][y_train==1], X_train[:, 1][y_train==1], c='g', marker='^', s=30)
plt.scatter(X_train[:, 0][y_train==2], X_train[:, 1][y_train==2], c='y', marker='s', s=30)
plt.contourf(x0, x1, u, c=u, alpha=0.2)
plt.show()
forest = RandomForestClassifier(max_depth=5, num_trees=100, feature_sample_rate=0.5, data_sample_rate=0.15)
forest.fit(X_train, convert_to_vector(y_train))
y_pred = forest.predict(X_test)
print("random forest accuracy= {}".format(accuracy_score(y_test, y_pred)))plt.figure(2)
u= forest.predict(W).reshape(x0.shape)
plt.axis([-1.5, 2.5, -0.75, 1.25 ])
plt.scatter(X_train[:, 0][y_train==0], X_train[:, 1][y_train==0], c='b', marker='o', s=30)
plt.scatter(X_train[:, 0][y_train==1], X_train[:, 1][y_train==1], c='g', marker='^', s=30)
plt.scatter(X_train[:, 0][y_train==2], X_train[:, 1][y_train==2], c='y', marker='s', s=30)
plt.contourf(x0, x1, u, c=u, alpha=0.2)plt.show()

randomForestClassifier.py

import numpy as np
from machine_learning.homework.week10.decisionTreeClassifier import DecisionTreeClassifierclass RandomForestClassifier:def __init__(self, num_trees, max_depth, feature_sample_rate,data_sample_rate, random_state=0):self.max_depth, self.num_trees = max_depth, num_treesself.feature_sample_rate = feature_sample_rateself.data_sample_rate = data_sample_rateself.trees = []np.random.seed(random_state)def get_data_samples(self, X, y):shuffled_indices = np.random.permutation(len(X))size = int(self.data_sample_rate * len(X))selected_indices = shuffled_indices[:size]return X[selected_indices], y[selected_indices]def fit(self, X, y):for t in range(self.num_trees):X_t, y_t = self.get_data_samples(X, y)model = DecisionTreeClassifier(max_depth=self.max_depth,feature_sample_rate=self.feature_sample_rate)model.fit(X_t, y_t)self.trees.append(model)def predict_proba(self, X):probas = np.array([tree.predict_proba(X) for tree in self.trees])return np.average(probas, axis=0)def predict(self, X):proba = self.predict_proba(X)return np.argmax(proba, axis=1)

TreeNode.py

# 树节点
class Node:j = Nonetheta = Nonep = Noneleft = Noneright = None