0

我计划对组间的多个变量进行 kruskal-wallis 检验（例如，out1、out2 和 out3 在序数水平上的疾病差异），并根据递减的 KW 测试结果获得排序的变量 [例如 out2、out3、out1]。但是，收到错误消息：“ValueError: If using all scalar values, you must pass an index”。那么如何解决呢？

from scipy import stats
import pandas as pd
import numpy as np

data = pd.DataFrame({ 'ID': list(range(1,21)), 
                  'Disease': ["Cancer","Healthy","Adenoma","Cancer","Cancer","Healthy","Cancer","Healthy","Cancer","Cancer","Adenoma","Cancer","Healthy","Cancer","Healthy","Healthy","Healthy","Inflammation","Healthy","Inflammation"],
                   'Out1': np.random.randn(20),
                  'Out2': np.random.randn(20),
                  'Out3': np.random.randn(20)})

def KW_test(y, x):
  y = np.array(y)
  label, idx = np.unique(list(x), return_inverse=True)
  groups = [y[idx == i] for i, l in enumerate(label)]
  H, p = stats.kruskal(*groups)
  return H

results=[]
outcomes= list(data.columns)[2:]
for i in outcomes: 
   y=data[i]
   x=data.iloc[:, 1]
   KW_test=KW_test(y,x)
   results=pd.DataFrame({'outcomes': i, 'KW_test': KW_test}, columns=['outcomes', 'KW_test']) 
   results.append(results)
   # sorted the results decreasingly using KW_test results
   results=results.sort_value(["KW_test"], ascending=False)
   # return the sorted outcomes 
   print(results[outcomes])
4

1 回答 1

0

解决方案是使用“枚举”功能添加索引

for index, i in enumerate(features): 
    y=data[i]
    x=data.iloc[:, 4]
    KW_test_result=KW_test(y,x)
    if index == 0:
    results = pd.DataFrame({'features': [i], 'KW_test': [KW_test_result]}, columns=['features', 'KW_test']) 
    else: 
    new_result=pd.DataFrame({'features': [i], 'KW_test': [KW_test_result]}, columns=['features', 'KW_test']) 
    results = pd.concat([results, new_result])
 
results=results.sort_values(["KW_test"], ascending=False)
于 2021-12-22T00:06:19.900 回答