sklearn实战-乳腺癌细胞数据挖掘(博主自行录制视频) https://study.163.com/course/introduction.htm? courseid=1005269003 UTM _ campaign=commission UTM _ source=CP-40000000398149 UTM _ medium=share
医药统计项目联系QQ:231469242
3359 github.com/Thomas-haslwanter/stats intro _ python/tree/master/ISP/code _ quant lets/08 _ testsmeanvanvalue
#-- coding : utf-8-- *-importnumpyasnp # additionalpackagesfromscipy.stats.mstatsimportkruskalwallis ' '…current module 33603360 scipy.stats.mstatsthismodulis hmaskedarrays.mostofthesefunctionsaresimilartothoseinscipy.statsbutmighthavesmalldifficall esintheapiorinthealgorithmused. age,someapichangesarestillpossible.' ' # get the datacouldbeacomparisonofthesmoglevelsinfourdifferentcities 122] ) city 101、103、113、84 ] (city3=NP.array ([ 70,68,54,73,81,68 ] ) city4=NP.array ) 61、54、54 group3] defkruskawallis _ test (返回list true时显示有显著性差异,返回False时显示无显著性差异的print ' usekruskawallistest 3360 ' h, p=kruskalWallis(list_groups ) print'H value: ',h print ' p p # printtheresultsifp 0.05: print (thereisasignignifiction ) returntrueelse : print (' nosignificantdifferencebetweenthecities.' ) returnfalseKruskawallis_test (列表_ groups )
样本数据不是正态分布,两组数比较时用mann-whitney检验,三组以上时用kruskal-wallis检验
kruskal-wallis是独立的单因素方差检验的版本
kruskal-wallis可用于排序计算
示例数据
过程
H0和H1假设
自由度:组数-1,这里有三个组,自由度为3-=2
自由度为2,a=0.05,对应于键值5.99,如果计算出的值大于5.99,则拒绝原假设
对数据进行排序,并在表中填写相应的排序顺序
计算公式:
t是一系列排序之和
n是一组个数
的h值2.854小于5.99,不拒绝原假设
python风控记分卡建模与风控常识https://study.163.com/course/introduction.htm? courseid=1005214003 UTM _ campaign=commission UTM _ source=CP-400000000398149 UTM _ medium=share