如何使用混合了两个变量的自定义聚合函数创建 Pandas 交叉表？

Question

Suppose I have 4 variables A , B , C , D , I want the crosstab to look like this:

+-------+-----------------+-----------------+
|  A/B  |      Cat1       |      Cat2       |
+-------+-----------------+-----------------+
| Cat-a | Sum(C)/Count(D) | Sum(C)/Count(D) |
| Cat-b | Sum(C)/Count(D) | Sum(C)/Count(D) |
+-------+-----------------+-----------------+

Eg. My data looks like this:

+------+--------+--------+--------+
| Type | Gender | Height | Weight |
+------+--------+--------+--------+
| Dog  | F      |     80 |     60 |
| Dog  | F      |     75 |     57 |
| Dog  | M      |     90 |     68 |
| Cat  | F      |     50 |     50 |
| Cat  | F      |     53 |     53 |
| Cat  | M      |     56 |     55 |
| Cat  | M      |     60 |     54 |
| Cat  | M      |     65 |     60 |
+------+--------+--------+--------+

Now suppose I want the aggregate to be sum(weight)/max(height). The crosstab would look something like:

+-------------+------------+------------+
| Type/Gender |     M      |     F      |
+-------------+------------+------------+
| Cat         | 169/65=2.6 | 103/53=1.9 |
| Dog         | 68/90=0.75 | 117/80=1.4 |
+-------------+------------+------------+

Answer 1

First aggregate by GroupBy.agg with max and sum , then create new column by DataFrame.assign with division, reshape by Series.unstack and last data cleaning - DataFrame.reset_index with DataFrame.rename_axis :

df1 = (df.groupby(['Type','Gender'])
         .agg({'Height':'max','Weight':'sum'})
         .assign(New = lambda x: x.Weight / x.Height)['New']
         .unstack()
         .reset_index()
         .rename_axis(None, axis=1))

print (df1)
  Type         F         M
0  Cat  1.943396  2.600000
1  Dog  1.462500  0.755556