将数据框列与列表值匹配并将数据框附加到匹配行

Question

我在两个数据框中读取了两个不同的 csv。我想将列 df1['building_type] 与 df2['model'] 匹配并将相应的行附加到 df1.

数据框 1：

data = [{'length': '34', 'width': '58.5', 'height': '60.2', 'building_type': ['concrete','wood','steel','laminate']},
       {'length': '42', 'width': '33', 'height': '23', 'building_type': ['concrete_double','wood_double','steel_double']}]
df1 = pd.DataFrame(data)

print(df1)

数据框 2：


data2 = [{'type': 'A1', 'floor': '2', 'model': ['wood','laminate','concrete','steel']},
       {'type': 'B3', 'floor': '4',  'model': ['wood_double','concrete_double','steel_double']}]
df2=pd.DataFrame(data2)
print(df2)

最终数据帧：

   length   width   height  building_type                                 type  floor
0   34      58.5    60.2   [concrete, wood, steel, laminate]              A1    2
1   42      33      23     [concrete_double, wood_double, steel_double]   B3    4

Answer 1

pd.merge 似乎是必要的工具，但我们需要一个不可变的数据类型。 list 是可变的，无法加入。我们可以将list（可变）转换为tuple或frozenset，两者都是不可变的，可以用来连接。由于示例输出显示顺序无关紧要，因此我选择了 frozenset.

代码如下：

import pandas as pd

data = [{'length': '34', 'width': '58.5', 'height': '60.2', 'building_type': ['concrete','wood','steel','laminate']},
       {'length': '42', 'width': '33', 'height': '23', 'building_type': ['concrete_double','wood_double','steel_double']}]
df1 = pd.DataFrame(data)
print(df1)

data2 = [{'type': 'A1', 'floor': '2', 'model': ['wood','laminate','concrete','steel']},
       {'type': 'B3', 'floor': '4',  'model': ['wood_double','concrete_double','steel_double']}]
df2=pd.DataFrame(data2)
print(df2)


# Note: Merge fails on mutable dtype
# pd.merge(df1, df2, left_on='building_type', right_on='model')
# Produces `TypeError: unhashable type: 'list'`

# Convert mutable type to immutable type and merge.
# `tuple` is best if order matters for you. I am assuming that the
# order doesn't matter based on the sample output, so `frozenset` is more
# appropriate.
df1['building_type'] = df1['building_type'].apply(frozenset)
df2['model'] = df2['model'].apply(frozenset)

# Now, merge. Note that since column names are different both
# 'building_type' and 'model' would be retained. You can remove one of them.
final_df = pd.merge(df1, df2, left_on='building_type', right_on='model')
final_df = final_df.drop(['model'], axis=1)
print(final_df)

我机器上的输出：

  length width height                                 building_type
0     34  58.5   60.2             [concrete, wood, steel, laminate]
1     42    33     23  [concrete_double, wood_double, steel_double]
  type floor                                         model
0   A1     2             [wood, laminate, concrete, steel]
1   B3     4  [wood_double, concrete_double, steel_double]
  length width height                                 building_type type floor
0     34  58.5   60.2             (laminate, wood, steel, concrete)   A1     2
1     42    33     23  (concrete_double, steel_double, wood_double)   B3     4

将数据框列与列表值匹配并将数据框附加到匹配行

match the dataframe columns with list values and append dataframe with matching rows

python

pandas

dataframe

merge

concat