我正在尝试使用元组列表来重新索引多索引Pandas.Series。该系列的结构如下
(Pdb) object
respondent brand
0 Asda 6
1 Tesco 7
2 Asda 9
3 Aldi 2
4 Asda 4
Name: rating, dtype: int64
和多索引结构如下
(Pdb) obj.index
MultiIndex(levels=[[0, 1, 2, 3, 4], [u'Aldi', u'Asda', u'Tesco']],
labels=[[0, 1, 2, 3, 4], [1, 2, 1, 0, 1]],
names=[u'respondent', u'brand'])
按照以下步骤进行重新索引
indexes = [(0, u'Asda'), (0, u'Tesco'), (0, u'Aldi'), (0, u'pick'), (1, u'Asda'), (1, u'Tesco'), (1, u'Aldi'), (1, u'pick'), (2, u'Asda'), (2, u'Tesco'), (2, u'Aldi'), (2, u'pick'), (3, u'Asda'), (3, u'Tesco'), (3, u'Aldi'), (3, u'pick'), (4, u'Asda'), (4, u'Tesco'), (4, u'Aldi'), (4, u'pick')]
obj.reindex(index=indexes, fill_value=default)
结果是
0 Asda 6
Tesco 0
Aldi 0
pick 0
1 Asda 0
Tesco 7
Aldi 0
pick 0
2 Asda 9
Tesco 0
Aldi 0
pick 0
3 Asda 0
Tesco 0
Aldi 2
pick 0
4 Asda 4
Tesco 0
Aldi 0
pick 0
,多索引现在是
MultiIndex(levels=[[0, 1, 2, 3, 4], [u'Aldi', u'Asda', u'Tesco', u'pick']],
labels=[[0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4], [1, 2, 0, 3, 1, 2, 0, 3, 1, 2, 0, 3, 1, 2, 0, 3, 1, 2, 0, 3]])
这是一个问题,因为索引名称已被删除。有什么方法可以防止这种情况吗?如何解决?
答案 0 :(得分:1)
我想我们可以制作新的DataFrame
并将值添加到原始DataFrame
中的新值:
In [134]:
print df
v
respondent brand
0 Asda 6
1 Tesco 7
2 Asda 9
3 Aldi 2
4 Asda 4
In [135]:
Index = pd.MultiIndex.from_product((df.index.get_level_values(0).unique(),
df.index.get_level_values(1).unique()),
names = df.index.names)
df2 = pd.DataFrame({'v': np.zeros(len(Index))},
index = Index)
In [136]:
print (df2+df).fillna(0)
v
respondent brand
0 Aldi 0
Asda 6
Tesco 0
1 Aldi 0
Asda 0
Tesco 7
2 Aldi 0
Asda 9
Tesco 0
3 Aldi 2
Asda 0
Tesco 0
4 Aldi 0
Asda 4
Tesco 0
答案 1 :(得分:0)
我最终采取了一种稍微不同的方法,这对我来说效果更好。我最终构建了一个新的Index / MultiIndex,正确命名然后对此进行重新索引。
def re_index(obj, default, indexes):
if isinstance(indexes[0], list):
complete_index = pd.MultiIndex.from_product(indexes)
complete_index.names = obj.index.names
else:
complete_index = pd.Index(data=indexes, name=obj.index.name)
complete_index.name = obj.index.name
new_obj = copy.copy(obj)
return new_obj.reindex(index=complete_index, fill_value=default)