Question

我有两个数据框，df1包含Id和Items，该列表的表示如下所示。

ITEM ID -

[(itemid, weight, 3)]

使用以下数据框，我们需要使用项目数据框值将列表中每个值的itemID的权重相加。

ID - DF1

ID  ITEMS
    11  [(123, 2.12,3),(234, 1.2,3)]
    22  [(567, 2.3, 3),(245, 1.9,3)]
    33  [(999,4.5, 3),(222, 2.0,3)]
    44  [(223, 2.34,3),(234,3.5,3)]

ITEM - DF2

ITEMS   WEIGHT
    123 2.5
    234 1.8
    567 19
    245 3
    999 2
    222 2.9
    223 4.2

预期产出：

ID  Items
11  [(123, 2.12+2.5,3),(234, 1.2+1.8,3)]
22  [(567,4.2,3),(245,4.9,3)]
33  [(999, 6.5, 3), (222, 4.9,3)]
44  [(223, 6.54,3),(234,5.3,3)]

第一行显示为示例， [（123,2.12 + 2.5,3），（234,1.2 + 1.8,3）] 。

Answer 1

这是使用pd.Series.apply的一种方式：

import pandas as pd

df1 = pd.DataFrame({'ID': [11, 22, 33, 44],
                    'ITEMS': [[(123, 2.12,3),(234, 1.2,3)],
                              [(567, 2.3, 3),(245, 1.9,3)],
                              [(999,4.5, 3),(222, 2.0,3)],
                              [(223, 2.34,3),(234,3.5,3)]]})

df2 = pd.DataFrame({'ITEMS': [123, 234, 567, 245, 999, 222, 223],
                    'WEIGHT': [2.5, 1.8, 19, 3, 2, 2.9, 4.2]})

s = df2.set_index('ITEMS')['WEIGHT']

df1['ITEMS'] = df1['ITEMS'].apply(lambda x: [(i[0], i[1]+s.get(i[0]), i[2]) for i in x])

print(df1)

#    ID                            ITEMS
# 0  11  [(123, 4.62, 3), (234, 3.0, 3)]
# 1  22  [(567, 21.3, 3), (245, 4.9, 3)]
# 2  33   [(999, 6.5, 3), (222, 4.9, 3)]
# 3  44  [(223, 6.54, 3), (234, 5.3, 3)]

在我看来，如果可能的话，最好将数字数据分成不同的列并使用矢量化功能。

Answer 2

这是另一个更新行的方法。

import pandas as pd
import numpy as np
items = {'id': [11, 22, 33, 44], 'items': [[(123, 2.12,3),(234, 1.2,3)],
                                      [(567, 2.3, 3),(245, 1.9,3)],
                                      [(999,4.5, 3),(222, 2.0,3)],
                                      [(223, 2.34,3),(234,3.5,3)]
                                      ]}

df1 = pd.DataFrame(data=items)
item_weight_data = {'items': [123, 234, 567, 245, 999, 222, 223], 'weight':[2.5, 1.8, 19, 3, 2, 2.9, 4.2]}
df2 = pd.DataFrame(data=item_weight_data)
df2 = df2.set_index('items')


#function that takes row and dataframe as input and returns new row.
def update_weight(row, item_df):
    try:
        new_row = [];
        for item in row:

            weight = item_df.loc[item[0],'weight']
            #since item is a tuple, It cannot be updated.
            #so creating new updated tuple and appending it to the list.
            updated_item =  (item[0],(item[1] + weight),item[2])
            new_row.append(updated_item)

        return new_row
    except Exception as e:
        raise ValueError("UNEXPECTED_DATA")



df1['items'] = df1['items'].apply(lambda x: update_weight(x, df2))
print(df1)

我希望它有所帮助。

映射两个数据帧的列并从列表中添加值

2 个答案: