假设检验

import numpy as np

import pandas as pd

调试代码的数据集：dir(datasets)

# 加载数据（鸢尾花）

from sklearn import datasets

dir(datasets)

iris_raw=datasets.load_iris()

iris_raw

iris_raw.target # target是分类,调出

iris_raw.data

# array转dataframe

pd.DataFrame(iris_raw.data)

# 重新命名列名

# array转dataframe

iris=pd.DataFrame(iris_raw.data,columns=iris_raw.feature_names)

iris

# pd.DataFrame(iris_raw.data,columns=['a','b','c','d'])

# 增加一列

iris['Species']=iris_raw.target # target原本是array

iris

正态检验

iris['petal length (cm)'].mean()

# H0：鸢尾花花瓣平均长度是4.2

import statsmodels.stats.weightstats as sw

sw.ztest(iris['petal length (cm)'],value=4.2)

#(-3.066548320028344, 0.0021654580512200875) 分别对应统计值和P值

t检验

# H0: 花瓣平均长度为4.0

import scipy.stats as ss

ss.ttest_1samp(iris['petal length (cm)'],popmean=4.0)

# 取出数值

stats_val,p_val=ss.ttest_1samp(iris['petal length (cm)'],popmean=4.0)

print(stats_val,p_val)

双样本检验

iris['Species'].unique()

# 山鸢尾和可变色鸢尾，花瓣长度是否有差异

iris['petal length (cm)']

import scipy.stats as ss

ss.ttest_ind(iris[iris['Species']==0]['petal length (cm)'], iris[iris['Species']==1]['petal length (cm)'])

# Pvalue小于0.025，拒绝原假设