决策树

%matplotlib inline
import matplotlib.pyplot as plt
import pandas as pd

from sklearn.datasets.california_housing import fetch_california_housing
housing = fetch_california_housing() # 内置的数据集

housing.data.shape # (20640,8)

from sklearn import tree
dtr = tree.DecisionTreeRegressor(max_depth = 2) # 构造决策树模型，并指定最大的深度为2
dtr.fit(housing.data[:,[6,7]],housing.target)

可视化显示(要先安装一个模块）
dot_data = tree.export_graphviz(dtr,out_file=None,feature_names=housing.feature_names[6:8],filled=True,impurity=False,rounded=True)
# feature_names 特征名字

# 要先安装
import pydotplus
graph = pydotplus.graph_from_dot_data(dot_data)
graph.get_nodes()[7].set_fillcolor("#FFF2DD") # 满足第七个节点为True的时候显示的颜色
from IPython.display import Image
Image(graph.create_png()) 
graph.write_png("dtr_white_background.png")

from sklearn.model_selection import train_test_split
data_train,data_test,target_train,target_test = train_test_split(housing.data,housing.target,test_size=0.1,random_state=42) # random_state每次随机的测试集
# 都是一样的，它指的是以中状态
dtr = tree.DecisionTreeRegressor(random_state=42)
dtr.fit(data_train,target_train)
dtr.score(data_test,target_test)

from sklearn.grid_search import GridSearchCV
tree_param_grid = {'min_samples_split':list((3,6,9)),'n_estimators':list((10,50,100))}
# min_samples_split:叶子节点至少为多少的时候才能进行分裂，n_estimators:?
grid = GridSearchCV(RandomForestRegressor(),param_grid=tree_param_grid,cv=5) # cv交叉验证
grid.fit(dat_train,target_train)
grid.grid_scores_,grid.best_param_,grid.best_score_

 

 

转载于:https://www.cnblogs.com/panjie123pakho/p/7876020.html