熊猫-枢轴多个分类列

Question

I have a dataframe as such:

name = ['fred','fred','fred','james','james','rick','rick','jeff']
actionfigures = ['superman','batman','flash','greenlantern','flash','batman','joker','superman']
cars = ['lamborghini', 'ferrari','bugatti','ferrari','corvette','bugatti','bmw','bmw']
pets = ['cat','dog','bird','cat','dog','dog','fish','marmet']

test = pd.DataFrame({'name':name,'actfig':actionfigures,'car':cars,'pet':pets})

    actfig       car                name    pet
0   superman     lamborghini        fred    cat
1   batman       ferrari            fred    dog
2   flash        bugatti            fred    bird
3   greenlantern ferrari            james   cat
4   flash        corvette           james   dog
5   batman       bugatti            rick    dog
6   joker        bmw                rick    fish
7   superman     bmw                jeff    marmet

Forgive me if my terminology is incorrect, but I want to pivot the data so that I get counts for each value in the ['actionfigures','car','pet'] columns for each name.

    batman  flash   greenlantern    joker   superman    bmw bugatti corvette    ferrari lamborghini bird    cat dog fish    marmet
name                                                            
fred    1   1   0   0   1   0   1   0   1   1   1   1   1   0   0
james   0   1   1   0   0   0   0   1   1   0   0   1   1   0   0
jeff    0   0   0   0   1   1   0   0   0   0   0   0   0   0   1
rick    1   0   0   1   0   1   1   0   0   0   0   0   1   1   0

I would have thought that test.pivot_table(index='name',columns=['actfig','car','pet'],aggfunc='size']) would do it, but it gives me some weird multi-level columns.

Thinking maybe I could concat get_dummies for each column then groupby name and sum, but feel like pandas prob has a better way.

How would this be done?

Answer 1

melt and pivot

test.melt('name').assign(new=1).pivot('name','value','new').fillna(0)
Out[239]: 
value  batman  bird  bmw  bugatti  cat  corvette  dog  ferrari  fish  flash  \
name                                                                          
fred      1.0   1.0  0.0      1.0  1.0       0.0  1.0      1.0   0.0    1.0   
james     0.0   0.0  0.0      0.0  1.0       1.0  1.0      1.0   0.0    1.0   
jeff      0.0   0.0  1.0      0.0  0.0       0.0  0.0      0.0   0.0    0.0   
rick      1.0   0.0  1.0      1.0  0.0       0.0  1.0      0.0   1.0    0.0   
value  greenlantern  joker  lamborghini  marmet  superman  
name                                                       
fred            0.0    0.0          1.0     0.0       1.0  
james           1.0    0.0          0.0     0.0       0.0  
jeff            0.0    0.0          0.0     1.0       1.0  
rick            0.0    1.0          0.0     0.0       0.0  

Or get_dummies

pd.get_dummies(test.set_index('name')).sum(level=0)
Out[248]: 
       actfig_batman  actfig_flash  actfig_greenlantern  actfig_joker  \
name                                                                    
fred               1             1                    0             0   
james              0             1                    1             0   
jeff               0             0                    0             0   
rick               1             0                    0             1   
       actfig_superman  car_bmw  car_bugatti  car_corvette  car_ferrari  \
name                                                                      
fred                 1        0            1             0            1   
james                0        0            0             1            1   
jeff                 1        1            0             0            0   
rick                 0        1            1             0            0   
       car_lamborghini  pet_bird  pet_cat  pet_dog  pet_fish  pet_marmet  
name                                                                      
fred                 1         1        1        1         0           0  
james                0         0        1        1         0           0  
jeff                 0         0        0        0         0           1  
rick                 0         0        0        1         1           0

Edit: As per PiR

pd.get_dummies(test.set_index('name'), prefix_sep='|').sum(level=0).rename(columns=lambda c: c.rsplit('|', 1)[1]) 

Answer 2

Option 1
pd.get_dummies by parts

a = pd.get_dummies(test.actfig)
c = pd.get_dummies(test.car)
p = pd.get_dummies(test.pet)
n = pd.get_dummies(test.name).T

pd.concat([n.dot(d) for d in [a, c, p]], axis=1)

       batman  flash  greenlantern  joker  superman  bmw  bugatti  corvette  ferrari  lamborghini  bird  cat  dog  fish  marmet
fred        1      1             0      0         1    0        1         0        1            1     1    1    1     0       0
james       0      1             1      0         0    0        0         1        1            0     0    1    1     0       0
jeff        0      0             0      0         1    1        0         0        0            0     0    0    0     0       1
rick        1      0             0      1         0    1        1         0        0            0     0    0    1     1       0

---

Option 2
stack + pd.crosstab

test.set_index('name').stack().pipe(
    lambda x: pd.crosstab(x.index.get_level_values(0), x.values))

col_0  batman  bird  bmw  bugatti  cat  corvette  dog  ferrari  fish  flash  greenlantern  joker  lamborghini  marmet  superman
row_0                                                                                                                          
fred        1     1    0        1    1         0    1        1     0      1             0      0            1       0         1
james       0     0    0        0    1         1    1        1     0      1             1      0            0       0         0
jeff        0     0    1        0    0         0    0        0     0      0             0      0            0       1         1
rick        1     0    1        1    0         0    1        0     1      0             0      1            0       0         0