如何计算groupby数据框中列的值？

Question

So I'm having trouble with this one:

Ive tried this, but it stacks time:

df2 = df.groupby(['Time'])
for group, data in df2:
    result =  data.apply(lambda x: x.value_counts()).T.stack()
    print(result)

Answer 1

You are trying to use the values of Time as the new columns and the values of the other columns as new index. If you had just a few columns, this could be easily achieved by a pivot_table . Eg for 'Health' :

In [2]: df = pd.DataFrame([['T0', 'Yes', 'Good'], ['T0', 'Yes', 'Bad'], ['T1', 'No', 'Good'], ['T1', 'No', 'Good']], columns=['Time', 'Health', 'Meds'])    

In [18]: pd.pivot_table(df[['Health', 'Time']], index='Health', columns='Time', aggfunc='size', fill_value=0)                                                                                                      
Out[18]: 
Time    T0  T1
Health        
No       0   2
Yes      2   0

However, you want to repeat that procedure for all columns. This is also possible with a pivot table, given you reshape your dataframe into a long data format. This means you create a new column with all column names and is exactly what the stack() function is for:

In [45]: df_stacked = df.set_index('Time').stack().rename('value').reset_index()                                                                                                                                                
Out[45]: 
  Time level_1 value
0   T0  Health   Yes
1   T0    Meds  Good
2   T0  Health   Yes
3   T0    Meds   Bad
4   T1  Health    No
5   T1    Meds  Good
6   T1  Health    No
7   T1    Meds  Good

Now you can pivot the stacked dataframe in order to use both, the values of the column with all column names and their values as index

In [48]: pd.pivot_table(df_stacked, index=['level_1', 'value'], columns='Time', aggfunc='size', fill_value=0)                                                                                                      
Out[48]: 
Time           T0  T1
level_1 value        
Health  No      0   2
        Yes     2   0
Meds    Bad     1   0
        Good    1   2

The description is probably a bit confusing but I hope the code makes it clear. You basically had the right ingredients but the combination is a bit tricky.