calculate a median value of pd.DataFrame() index based on values in the column

Question

Lets cay I have a pd.DataFrame() object that stores number of people that given age ang gender had stroke in the past. In mor visual way:

positive_by_gender.tail()

gives us:

gender	Female	Male
age
78	9.0	12.0
79	13.0	4.0
80	10.0	7.0
81	8.0	6.0
82	4.0	5.0

So there are 9 females of age 78 that had stroke, 12 males of age 78 that had stroke etc.

What I want is to calculate a median for each gender of age taht they had stroke - in this excample it would be 79.5 for females, but I want it to be calculated by code not by me:-) - I guess I could make an array that for females would look like: [78 times 9, 79 times 13, 80 times 10, etc... ] and then find median this way but still - I dunno how to do even that. I'd really appreciate all help.

Answer 1

To follow your idea of creating an array and getting the median this way:

In [235]: df
Out[235]: 
     Female  Male
age              
78      9.0  12.0
79     13.0   4.0
80     10.0   7.0
81      8.0   6.0
82      4.0   5.0

In [236]: df = df.astype(int)

In [237]: df
Out[237]: 
     Female  Male
age              
78        9    12
79       13     4
80       10     7
81        8     6
82        4     5

In [238]: df = df.reset_index('age')

In [240]: df = df.melt(id_vars='age', var_name='gender', value_name='count')

In [241]: df
Out[241]: 
   age  gender  count
0   78  Female      9
1   79  Female     13
2   80  Female     10
3   81  Female      8
4   82  Female      4
5   78    Male     12
6   79    Male      4
7   80    Male      7
8   81    Male      6
9   82    Male      5

In [242]: df['age'] = df.apply(lambda s: [s['age']] * s['count'], axis=1)

In [243]: df
Out[243]: 
                                                 age  gender  count
0               [78, 78, 78, 78, 78, 78, 78, 78, 78]  Female      9
1  [79, 79, 79, 79, 79, 79, 79, 79, 79, 79, 79, 7...  Female     13
2           [80, 80, 80, 80, 80, 80, 80, 80, 80, 80]  Female     10
3                   [81, 81, 81, 81, 81, 81, 81, 81]  Female      8
4                                   [82, 82, 82, 82]  Female      4
5   [78, 78, 78, 78, 78, 78, 78, 78, 78, 78, 78, 78]    Male     12
6                                   [79, 79, 79, 79]    Male      4
7                       [80, 80, 80, 80, 80, 80, 80]    Male      7
8                           [81, 81, 81, 81, 81, 81]    Male      6
9                               [82, 82, 82, 82, 82]    Male      5

In [245]: df = df.explode('age')
In [249]: df['age'] = df['age'].astype(int)

In [251]: df
Out[251]: 
    age  gender  count
0    78  Female      9
0    78  Female      9
0    78  Female      9
0    78  Female      9
0    78  Female      9
..  ...     ...    ...
9    82    Male      5
9    82    Male      5
9    82    Male      5
9    82    Male      5
9    82    Male      5

[78 rows x 3 columns]

In [250]: df.groupby('gender')['age'].median()
Out[250]: 
gender
Female    79.5
Male      80.0
Name: age, dtype: float64