填充多索引熊猫系列

Question

我有一个充满数据的 pandas 数据框

import pandas as pd
import numpy as np

varNames = ["point1","point2","point3","point4","point5"]
df = pd.DataFrame(np.random.randn(5,2),index=varNames,columns=["data1","data2"])

我想创建一个带有由此创建的 multiIndex 的系列。我能做的索引：

iterables=[["point1","point2","point3"],["point4","point5"]]
index=pd.MultiIndex.from_product(iterables, names=['numerator', 'denominator'])

虽然我不知道如何填写这个系列。我在追求类似的东西

s = pd.Series(max(df.loc[index["numerator"]]/df.loc[index["denominator"]]),index=index)

我想将第一个数据帧中列为分子的每一行除以第一个数据帧中列为分母的每一行，从结果值行中找到最大值并将其存储在相关的将 (s[variableN,variableM]) 放在系列中。

这是我第一次使用这个多索引的东西，没有逐行遍历系列，计算出值并存储它，类似（我想，我不认为我能够完全理解这一点）this，我不知道该怎么做。

Answer 1

您可以将reindex 与参数level 与max 一起使用：

df3 = df.reindex(index, level=0).div(df.reindex(index, level=1)).max(level=0)

示例：

np.random.seed(456)
varNames = ["point1","point2","point3","point4","point5"]
df = pd.DataFrame(np.random.randn(5,2),index=varNames,columns=["data1","data2"])
print (df)
           data1     data2
point1 -0.668129 -0.498210
point2  0.618576  0.568692
point3  1.350509  1.629589
point4  0.301966  0.449483
point5 -0.345811 -0.315231

iterables=[["point1","point2","point3"],["point4","point5"]]
index=pd.MultiIndex.from_product(iterables, names=['numerator', 'denominator'])

---

df1 = df.reindex(index, level=0)
print (df1)
                          data1     data2
numerator denominator                    
point1    point4      -0.668129 -0.498210
          point5      -0.668129 -0.498210
point2    point4       0.618576  0.568692
          point5       0.618576  0.568692
point3    point4       1.350509  1.629589
          point5       1.350509  1.629589

df2 = df.reindex(index, level=1)
print (df2)
                          data1     data2
numerator denominator                    
point1    point4       0.301966  0.449483
          point5      -0.345811 -0.315231
point2    point4       0.301966  0.449483
          point5      -0.345811 -0.315231
point3    point4       0.301966  0.449483
          point5      -0.345811 -0.315231

print (df1.div(df2))
                          data1     data2
numerator denominator                    
point1    point4      -2.212594 -1.108405
          point5       1.932062  1.580459
point2    point4       2.048493  1.265214
          point5      -1.788768 -1.804050
point3    point4       4.472386  3.625472
          point5      -3.905339 -5.169509

---

df3 = df.reindex(index, level=0).div(df.reindex(index, level=1)).max(level=0)
print (df3)
              data1     data2
numerator                    
point1     1.932062  1.580459
point2     2.048493  1.265214
point3     4.472386  3.625472


df3 = (df.reindex(index, level=0).div(df.reindex(index, level=1))
        .max(level=0)
        .reindex(index, level=0))
print (df3)
                          data1     data2
numerator denominator                    
point1    point4       1.932062  1.580459
          point5       1.932062  1.580459
point2    point4       2.048493  1.265214
          point5       2.048493  1.265214
point3    point4       4.472386  3.625472
          point5       4.472386  3.625472