python - 如何使用 Python pandas 将某些列扩展为行

Question

我有一个包含多对多关系的两列的表。

例如

Animal     Food
rabbit     grass
rabbit     carrots
rabbit     cabbage
dog        carrots
horse      grass
horse      hay

我想要这样的东西:

Animal     Food1   Food2     Food3 
rabbit     grass   carrots   cabbage  
dog        carrots
horse      grass   hay

或者类似这样的东西:

Animal     Grass     Carrots    Cabbage     Hay
rabbit      True      True       True       False
dog         False     True       False      False
horse       True      False      False      True

我已经尝试过数据透视表、熔化和堆叠，但仍然不知道如何执行此操作。任何帮助将不胜感激。谢谢!!

Answer 1

选项 1
groupby 与 size

df.groupby(['Animal', 'Food']).size().unstack(fill_value=0).astype(bool)

Food    cabbage  carrots  grass    hay
Animal                                
dog       False     True  False  False
horse     False    False   True   True
rabbit     True     True   True  False

---

s = df.groupby('Animal').Food.apply(list)
pd.DataFrame(s.values.tolist(), s.index).add_prefix('Food').fillna('')

          Food0    Food1    Food2
Animal                           
dog     carrots                  
horse     grass      hay         
rabbit    grass  carrots  cabbage

---

选项 2
groupby 和 value_counts

df.groupby('Animal').Food.value_counts().unstack(fill_value=0).astype(bool)

Food    cabbage  carrots  grass    hay
Animal                                
dog       False     True  False  False
horse     False    False   True   True
rabbit     True     True   True  False

---

选项 3
groupby 和 str.get_dummies

df.groupby('Animal').Food.apply('|'.join).str.get_dummies().astype(bool)

        cabbage  carrots  grass    hay
Animal                                
dog       False     True  False  False
horse     False    False   True   True
rabbit     True     True   True  False

---

选项 4
pandas.factorize 和 numpy.bincount

f1, u1 = pd.factorize(df.Animal.values)
f2, u2 = pd.factorize(df.Food.values)

n = u1.size
m = u2.size

b = np.bincount(f1 * m + f2, minlength=n * m).reshape(n, m)

pd.DataFrame(b.astype(bool), u1, u2)

        grass  carrots  cabbage    hay
rabbit   True     True     True  False
dog     False     True    False  False
horse    True    False    False   True

选项 5
无聊...所以想出了更多

f, u = pd.factorize(df.Animal.values)
n = u.size

a = [[] for _ in range(n)]
[a[i].append(food) for i, food in zip(f, df.Food)];
pd.DataFrame(a, u).rename(columns=lambda x: x+1).add_prefix('Food').fillna('')

          Food1    Food2    Food3
rabbit    grass  carrots  cabbage
dog     carrots                  
horse     grass      hay