画heatmaps取相关系数大的特征。

import pandas as pd

import seaborn as sns

import matplotlib.pyplot as plt

from scipy import stats

train_data='zhengqi_train.txt'

test_data='zhengqi_test.txt'

train=pd.read_csv(train_data,sep='\t',encoding='utf-8')

test=pd.read_csv(test_data,sep='\t',encoding='utf-8')

train.drop(['V5','V9','V11','V17','V22','V28'],axis=1)

co=train.corr()

print(co.nlargest(10,'target'))#输出与target相关系数大于10的特征。满足要求指定行和所有列。

print(co.nlargest(10,'target')['target'])#在上一条基础上取出target列

cols=co.nlargest(10,'target')['target'].index#在列上输出索引。

import numpy as np

sns.heatmap(train[cols].corr())#有了列输出特征。

plt.show()