用scikit-learn做特征选择(基于决策树)

导入相应模块

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from sklearn.tree import DecisionTreeRegressor

读取数据

scoring='neg_root_mean_squared_error'
X=pd.read_excel('boston.xlsx')
del X['Unnamed: 0']
Y=X['target'].copy()
del X['target']
names=X.columns

用决策树模型做特征选择

Tree = DecisionTreeRegressor()
Tree.fit(X,Y)

DecisionTreeRegressor()

print('feature'+'\t'+'importances')
for i in range(len(names)):
    print('%s\t%.5f'%(names[i],Tree.feature_importances_[i]))

feature	importances
CRIM	0.03738
ZN	0.00054
INDUS	0.00571
CHAS	0.00088
NOX	0.04893
RM	0.59258
AGE	0.00993
DIS	0.07267
RAD	0.00071
TAX	0.01198
PTRATIO	0.00837
B	0.01498
LSTAT	0.19533

排序后的特征

sortindex=np.argsort(Tree.feature_importances_)[::-1]
for i in sortindex:
    print('%s\t%.5f'%(names[i],Tree.feature_importances_[i]))

RM	0.59258
LSTAT	0.19533
DIS	0.07267
NOX	0.04893
CRIM	0.03738
B	0.01498
TAX	0.01198
AGE	0.00993
PTRATIO	0.00837
INDUS	0.00571
CHAS	0.00088
RAD	0.00071
ZN	0.00054