机器学习—决策树

Posted 2020-10-30 慢慢来会比较快

一、原理部分：

还是图片显示~

 

 

二、sklearn实现

1、回归树

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib as mpl
import seaborn as sns
mpl.rcParams[\'font.sans-serif\'] = [u\'SimHei\']
mpl.rcParams[\'axes.unicode_minus\'] = False
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error

#制造一些数据
x_data = np.random.rand(200,5)*10
w = np.array([2,4,6,8,10])
y = np.dot(x_data,w) + np.random.rand(200)*20 + 20
x_train,x_test,y_train,y_test = train_test_split(x_data,y)

#先试试决策回归树吧，都没试过
from sklearn.tree import DecisionTreeRegressor
dtr = DecisionTreeRegressor()
dtr.fit(x_train,y_train)
y_hat = dtr.predict(x_test)
print(\'决策回归树的误差：\',mean_squared_error(y_test,y_hat))

#画图
fig,ax = plt.subplots()
fig.set_size_inches(10,6)
ax.plot(np.arange(len(y_hat)),y_hat,color = \'r\')
ax.plot(np.arange(len(y_hat)),y_test,color = \'g\')

决策树做回归也太差了吧，难道是我调参有问题吗？一会试试调参看看

决策回归树的误差： 667.87208618

#调参一下试试
from sklearn.model_selection import GridSearchCV
model_dtr = GridSearchCV(dtr,param_grid=({\'max_depth\':np.arange(1,50)}),cv=10)
model_dtr.fit(x_train,y_train)
y_hat = model_dtr.predict(x_test)
print(model_dtr.best_params_)
print(\'决策回归树的误差：\',mean_squared_error(y_test,y_hat))

还是没啥用，好差的效果，同样的数据，前面线性回归的均方误差才二十几

决策回归树的误差： 643.989924585

 2、分类树

from sklearn.datasets import load_digits
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import accuracy_score
digits = load_digits()
x_data = digits.data
y_data = digits.target
x_train,x_test,y_train,y_test = train_test_split(x_data,y_data)
dtc = DecisionTreeClassifier()
dtc.fit(x_train,y_train)
y_hat = dtc.predict(x_test)
print(\'正确率\',accuracy_score(y_hat,y_test))


#调个参数看看
model_dtc = GridSearchCV(dtc,param_grid=({\'max_depth\':np.arange(1,20)}),cv=10)
model_dtc.fit(x_train,y_train)
print(\'最佳参数\',model_dtc.best_params_)
y_hat = model_dtc.predict(x_test)
y_hat = dtc.predict(x_test)
print(\'正确率\',accuracy_score(y_hat,y_test))

 

 正确率 0.817777777778

最佳参数 {\'max_depth\': 18}
正确率 0.817777777778

决策树还是弱分类器啊，难怪都喜欢用它来做ensemble