1、自定義函數(shù)

# 數(shù)據(jù)集 分組項(xiàng) 統(tǒng)計(jì)值
from statsmodels.stats.multicomp import pairwise_tukeyhsd
from statsmodels.stats.multicomp import MultiComparison

def pvaluegroup(data,groupcow,static):
    mc = MultiComparison(data[static], data[groupcow])
    result = mc.tukeyhsd()#計(jì)算各組間pvalue
    print(result.summary())
    #計(jì)算各組平均值，從高到低排序
    data_average=pd.DataFrame(data.groupby([groupcow])[static].mean())
    data_average=data_average.sort_values([static],ascending=False)
    data_average=data_average.reset_index(drop=False)
    data_average=data_average.astype(str)
    #修改格式
    dataresult=pd.DataFrame.from_records(result.summary())
    dataresult=dataresult.astype(str)
    # new_header = ["group1", "group2", "meandiff", "p-adj", "lower", "upper", "reject"]
    new_header = dataresult.iloc[0]
    dataresult.columns = new_header
    dataresult = dataresult[1:] # 刪除第0行
    # dataresult=dataresult.astype(str)
    dataresult["p-adj"]=dataresult["p-adj"].astype(float)
    #設(shè)置顯著性組
    data_average["顯著性組"]=""
    p_level_abc_num=0#角標(biāo)順序
    for i in range(len(data_average[groupcow])):
        p_level=[]
        #角標(biāo)字母
        p_level_abc=["a","b","c","d","e","f","g","h"]
    #     print(data_average.loc[i,groupcow])
        for j in range(i+1,len(data_average[groupcow])):
#             print(i,j,data_average.loc[i,groupcow],data_average.loc[j,groupcow])
            a=float(pd.concat([dataresult[(dataresult["group1"]==data_average.loc[i,groupcow])&(dataresult["group2"]==data_average.loc[j,groupcow])]["p-adj"],
                               dataresult[(dataresult["group1"]==data_average.loc[j,groupcow])&(dataresult["group2"]==data_average.loc[i,groupcow])]["p-adj"]]))
            if a<0.05 :#p=0.05
                p_level.append(data_average.loc[j,groupcow])
                #追加顯著性組
                data_average.loc[j,"顯著性組"]=str(data_average.loc[j,"顯著性組"])+p_level_abc[p_level_abc_num]
#                 print(p_level)
        if p_level != []:
            data_average.loc[i,"顯著性組"]=str(data_average.loc[i,"顯著性組"])+p_level_abc[i]
            p_level_abc_num+=1  #角標(biāo)移至下一位
        
    #     data_average.loc[i,"顯著性組"]=p_level
    data_average=data_average.sort_values([groupcow]) #按組別升序
    return data_average

數(shù)據(jù)集、數(shù)據(jù)分組列、數(shù)據(jù)值列，分組列可以為數(shù)字或字母，不同水平需合成一列，如：公母兩個(gè)性別、大白長(zhǎng)白兩個(gè)品種，需合并成一列，分別對(duì)應(yīng)大白公、大白母、長(zhǎng)白公、長(zhǎng)白母

輸出結(jié)果1

輸出結(jié)果2

2、匹配數(shù)據(jù)

df_out_group=pd.merge(a_30_pgrouop,a_100_pgrouop,left_on="品種性別",right_on="品種性別")
df_out_group=pd.merge(df_out_group,a_120_pgrouop,left_on="品種性別",right_on="品種性別")
df_out_group['品種性別']=df_out_group['品種性別'].replace({"1.0":'杜洛克公', "2.0":'杜洛克母', "3.0":'長(zhǎng)白公', "4.0":'長(zhǎng)白母', "5.0":'大白公', "6.0":'大白母'})
df_out_group

image.png