使用 pandas 索引作为字典键，根据匹配键用值填充字典

Question

我有一个test_df 组织如下：

[in]
# Use the arrays to create a dataframe
testing_df =pd.DataFrame(test_array,columns=['transaction_id','product_id'])

# Split the product_id's for the testing data
testing_df.set_index(['transaction_id'],inplace=True)

print(testing_df.head(n=5))

[out]
                     product_id
transaction_id                 
001                      (P01,)
002                  (P01, P02)
003             (P01, P02, P09)
004                  (P01, P03)
005             (P01, P03, P05)

然后我对其进行了一些计算并创建了一个字典来存储结果：

# Initialize a dictionary to store the matches
matches = {}

# Return the product combos values that are of the appropriate length and the strings match
for transaction_id,i in enumerate (testing_df['product_id']):
    recommendation = None
    recommended_count = 0

    for k, count in product_combos.items():
        k = list(k)
        if len(i)+1 == len(k) and count >= recommended_count:
            for product in i:
                if product in k: 
                    k.remove(product)
            if len(k) == 1:
                recommendation = k[0]
                recommended_count = count
    matches[transaction_id] = recommendation

print(matches)

[out]
{0: 'P09', 1: 'P09', 2: 'P06', 3: 'P09', 4: 'P09', 5: 'P09'}

我遇到的问题是 matches 字典的键应该是 001,002,003,004,005 等 - 对应于 test_df 的索引，即 001-100。

我遇到的第二个问题是我想用 001-100 的键填充另一个字典 (recommendations)。我想通过匹配键值将matches 中的值填充到这个字典中。

Answer 1

这里有几个问题。首先，您从enumerate 请求的变量的顺序被切换 - 整数计数器排在第一位：

for i, entry in enumerate(values):
    ...

这就是matches 字典中的键显示为整数的原因。

其次，您仍然需要访问testing_df.index 的ith 元素才能获得您正在寻找的transaction_id。您可以使用（更正后的）enumerate() 中的 i 执行此操作：

# sample data
transaction_id = ["001","002","003","004","005"]
product_id = {"product_id":[("P01",), ("P01", "P02"), ("P01", "P02", "P09"),
                            ("P01", "P03"), ("P01", "P03", "P05")]}
testing_df = pd.DataFrame(product_id, index=transaction_id)
testing_df.index.name = "transaction_id"

print(testing_df)
                     product_id
transaction_id                 
001                      (P01,)
002                  (P01, P02)
003             (P01, P02, P09)
004                  (P01, P03)
005             (P01, P03, P05)

matches = {}

for i, entry in enumerate(testing_df.product_id):

    # ... some computation ...

    transaction_id = testing_df.index[i]
    recommendation = entry[0] # just as an example
    matches[transaction_id] = recommendation

print(matches)
{'001': 'P01', '002': 'P01', '003': 'P01', '004': 'P01', '005': 'P01'}