python机器学习之决策树

利用决策树预测房价

# -*- coding: utf-8 -*-
from __future__ import unicode_literals
# 数据提供模块
import sklearn.datasets as sd
# 小工具模块
import sklearn.utils as su
import sklearn.tree as st
import sklearn.ensemble as se
# 评价模型指标模块
import sklearn.metrics as sm

# 获取数据
hosing = sd.load_boston()
print(hosing.data.shape)
# 房价
print(hosing.target.shape)
# 特征
print(hosing.feature_names)
# hosing.feature_names======》['CRIM'（犯罪情况） 'ZN'（房子密度） 'INDUS'（商铺繁华程度）
# 'CHAS'（是否靠河） 'NOX'（一氧化氮浓度） 'RM'(房间数) 'AGE'（年代） 'DIS'（离市中心的距离）
# 'RAD'（公路密度） 'TAX'（房产税） 'PTRATIO'（师生比例）'B'（黑人比例） 'LSTAT'(地位低下人比例)]
#'洗牌',打乱数据顺序，
x, y = su.shuffle(hosing.data, hosing.target, random_state=7)
# 确定训练集大小80%；
train_size = int(len(x) * 0.8)
# 分别拿出训练集和测试集
train_x, test_x, train_y, test_y = \
    x[: train_size], x[train_size: ], \
    y[: train_size], y[train_size:]
# 建立决策树回归器，最大高度为4层
model = st.DecisionTreeRegressor(max_depth=4)
# 构建决策树
model.fit(train_x, train_y)
# 生成预测集
pred_test_y = model.predict(test_x)
# 评判预测匹配度
print(sm.r2_score(test_y, pred_test_y))

# 建立正向激励回归器
model = se.AdaBoostRegressor(st.DecisionTreeRegressor(
    max_depth=4), n_estimators=400, random_state=7)
model.fit(train_x, train_y)
pred_test_y = model.predict(test_x)
# 评判预测匹配度
print(sm.r2_score(test_y, pred_test_y))

for test, pred_test in zip(test_y, pred_test_y):
    print(test, '->', pred_test)

特征值的重要性排序

　　不同的算法，对特征的认识不一样，现在对比一下决策树和正向激励的特征顺序。

# 评价模型指标模块
import sklearn.metrics as sm

import numpy as np
import sklearn.datasets as sd
import sklearn.utils as su
import sklearn.tree as st
import sklearn.ensemble as se
import matplotlib.pyplot as mp
hosing = sd.load_boston()
feature_names = hosing.feature_names
x, y = su.shuffle(hosing.data, hosing.target, random_state=7)
train_size = int(len(x) * 0.8)
train_x, test_x, train_y, test_y = \
    x[:train_size], x[train_size:], \
    y[:train_size], y[train_size:]
model = st.DecisionTreeRegressor(max_depth=4)
model.fit(train_x, train_y)
# 决策树特征值
feature_importances_dt = model.feature_importances_
print(feature_importances_dt)
model = se.AdaBoostRegressor(st.DecisionTreeRegressor(
    max_depth=4), n_estimators=400, random_state=7)
model.fit(train_x, train_y)
# 正向激励特征值
feature_importances_ab = model.feature_importances_
print(feature_importances_ab)

# 作图对比
mp.figure('Feature Importance', facecolor='lightgray')
# 决策树特征顺序
mp.subplot(211)
mp.title('Decision Tree', fontsize=16)
mp.ylabel('Importance', fontsize=12)
mp.tick_params(labelsize=10)
mp.grid(axis='y', linestyle=':')
# 逆向排序索引
sorted_indices = feature_importances_dt.argsort()[::-1]
pos = np.arange(sorted_indices.size)

mp.bar(pos, feature_importances_dt[sorted_indices],
       facecolor='deepskyblue', edgecolor='steelblue')
mp.xticks(pos, feature_names[sorted_indices], rotation=30)

# 正向激励决策树
mp.subplot(212)
mp.title('AdaBoost Decision Tree', fontsize=16)
mp.xlabel('Feature', fontsize=12)
mp.ylabel('Importance', fontsize=12)
mp.tick_params(labelsize=10)
mp.grid(axis='y', linestyle=':')
sorted_indices = feature_importances_ab.argsort()[::-1]
pos = np.arange(sorted_indices.size)
mp.bar(pos, feature_importances_ab[sorted_indices],
       facecolor='lightcoral', edgecolor='indianred')
mp.xticks(pos, feature_names[sorted_indices], rotation=30)

mp.tight_layout()
mp.show()

　　除了所选择的算法模型会对特征重要性的判断给出不同的排序以外，训练数据本身也会影响特征重要性的排列。

　

# -*- coding: utf-8 -*-
from __future__ import unicode_literals
import csv
import numpy as np
import sklearn.utils as su
import sklearn.ensemble as se
import sklearn.metrics as sm
import matplotlib.pyplot as mp
# 导入数据，单车以天为单位
with open('../ML/data/bike_day.csv', 'r') as f:
    reader = csv.reader(f)
    # x放特征，y放用户数
    x, y = [], []
    for row in reader:
                # 取前面的多个特征
        x.append(row[2:13])
        # y装用户数
        y.append(row[-1])
# 取出特征名
feature_names_dy = np.array(x[0])
# 取第一行后面的东西
x = np.array(x[1:], dtype=float)
y = np.array(y[1:], dtype=float)
x, y = su.shuffle(x, y, random_state=7)
# 取训练集90%
train_size = int(len(x) * 0.9)
train_x, test_x, train_y, test_y = \
    x[:train_size], x[train_size:], \
    y[:train_size], y[train_size:]
# 建立随机森林回归器
model = se.RandomForestRegressor(
    max_depth=10, n_estimators=1000, min_samples_split=2)
model.fit(train_x, train_y)
feature_importances_dy = model.feature_importances_
pred_test_y = model.predict(test_x)
print(sm.r2_score(test_y, pred_test_y))
# 导入数据，单车以小时为单位
with open('../ML/data/bike_hour.csv', 'r') as f:
    reader = csv.reader(f)
    x, y = [], []
    for row in reader:
        x.append(row[2:13])
        y.append(row[-1])
feature_names_hr = np.array(x[0])
x = np.array(x[1:], dtype=float)
y = np.array(y[1:], dtype=float)
x, y = su.shuffle(x, y, random_state=7)
train_size = int(len(x) * 0.9)
train_x, test_x, train_y, test_y = \
    x[:train_size], x[train_size:], \
    y[:train_size], y[train_size:]
model = se.RandomForestRegressor(
    max_depth=10, n_estimators=1000, min_samples_split=2)
model.fit(train_x, train_y)
feature_importances_hr = model.feature_importances_
pred_test_y = model.predict(test_x)
print(sm.r2_score(test_y, pred_test_y))
mp.figure('Feature Importance', facecolor='lightgray')
mp.subplot(211)
mp.title('Day', fontsize=16)
mp.ylabel('Importance', fontsize=12)
mp.tick_params(labelsize=10)
mp.grid(axis='y', linestyle=':')
sorted_indices = feature_importances_dy.argsort()[::-1]
pos = np.arange(sorted_indices.size)
mp.bar(pos, feature_importances_dy[sorted_indices],
       facecolor='deepskyblue', edgecolor='steelblue')
mp.xticks(pos, feature_names_dy[sorted_indices], rotation=30)
mp.subplot(212)
mp.title('Hour', fontsize=16)
mp.xlabel('Feature', fontsize=12)
mp.ylabel('Importance', fontsize=12)
mp.tick_params(labelsize=10)
mp.grid(axis='y', linestyle=':')
sorted_indices = feature_importances_hr.argsort()[::-1]
pos = np.arange(sorted_indices.size)
mp.bar(pos, feature_importances_hr[sorted_indices],
       facecolor='lightcoral', edgecolor='indianred')
mp.xticks(pos, feature_names_hr[sorted_indices], rotation=30)
mp.tight_layout()
mp.show()