单因素方差分析(spss单因素方差分析)

import numpy as np

import pandas as pd

import sklearn

import scipy.stats as stats

from statsmodels.formula.api import ols

from statsmodels.stats.anova import anova_lm

import warnings

warnings.filterwarnings(&34;ignore&34;)

dic_t1=[{&39;行业&39;:&39;A&39;,&39;次数&39;:57},

{&39;行业&39;:&39;A&39;,&39;次数&39;:66},

{&39;行业&39;:&39;A&39;,&39;次数&39;:49},

{&39;行业&39;:&39;A&39;,&39;次数&39;:40},

{&39;行业&39;:&39;A&39;,&39;次数&39;:34},

{&39;行业&39;:&39;A&39;,&39;次数&39;:53},

{&39;行业&39;:&39;A&39;,&39;次数&39;:44},

{&39;行业&39;:&39;B&39;,&39;次数&39;:68},

{&39;行业&39;:&39;B&39;,&39;次数&39;:39},

{&39;行业&39;:&39;B&39;,&39;次数&39;:29},

{&39;行业&39;:&39;B&39;,&39;次数&39;:45},

{&39;行业&39;:&39;B&39;,&39;次数&39;:56},

{&39;行业&39;:&39;B&39;,&39;次数&39;:51},

{&39;行业&39;:&39;C&39;,&39;次数&39;:31},

{&39;行业&39;:&39;C&39;,&39;次数&39;:49},

{&39;行业&39;:&39;C&39;,&39;次数&39;:21},

{&39;行业&39;:&39;C&39;,&39;次数&39;:34},

{&39;行业&39;:&39;C&39;,&39;次数&39;:40},

{&39;行业&39;:&39;D&39;,&39;次数&39;:44},

{&39;行业&39;:&39;D&39;,&39;次数&39;:51},

{&39;行业&39;:&39;D&39;,&39;次数&39;:65},

{&39;行业&39;:&39;D&39;,&39;次数&39;:77},

{&39;行业&39;:&39;D&39;,&39;次数&39;:58}]

df_t1 = pd.Dataframe(dic_t1,columns=[&39;行业&39;,&39;次数&39;])

def f_oneway(df,col_fac,col_sta):

list_fac = df[col_fac].unique() 分组标签取出

k = len(list_fac) 分组标签个数

n = len(df)

x_bar = df[col_sta].mean() 总均值

SSA = []

for i in list_fac:

series_i = df[df[col_fac]==i][col_sta]

r = len(series_i)

xi_bar = series_i.mean()

SSA.append(r*(xi_bar-x_bar)**2)

SST = round(((df[col_sta]-x_bar)**2).sum(),4)

df_res = pd.Dataframe(columns=[&39;方差来源&39;,&39;平方和&39;,&39;自由度&39;,&39;均方&39;,&39;F值&39;,&39;P-value&39;])

SSA = sum(SSA)

SSE = SST - SSA

MSA = SSA/(k-1)

MSE = SSE/(n-k)

sig = stats.f.sf(MSA/MSE,k-1,(k-1)*(n-1))

df_res[&39;方差来源&39;] = [col_fac,&39;组内(误差)&39;,&39;总和&39;]

df_res[&39;平方和&39;] = [SSA,SSE,SST]

df_res[&39;自由度&39;] = [k-1,n-k,n-1]

df_res[&39;均方&39;] = [MSA,MSE,&39;-&39;]

df_res[&39;F值&39;] = [MSA/MSE,&39;-&39;,&39;-&39;]

df_res[&39;P-value&39;] = [sig,&39;-&39;,&39;-&39;]

return df_res

f_oneway(df_t1,&39;行业&39;,&39;次数&39;)