根据条件从两个表创建一个新列

Question

medicament 有一个 id_questions_similar 的列表，每个问题在另一个 table(active_questions) 中都有一些 specialty_id。我需要收集所有 specialty_id 匹配 id_questions 并创建一个新列表，其中包含相应问题 ID 的所有专业 ID，并放置在 table medicament_similar

的新列中

data1 = {'id_questions': [1,2,3],
        'speciality_id': [[77, 66],[111, 122,133],[1245]]

        }

active_questions = pd.DataFrame(data1)
active_questions

data2 = {'id_medicament': [1, 2,3],
        'id_questions_similar': [[1, 2],[1,3], [3]]
        }

medicament_similar = pd.DataFrame(data2)

medicament_similar

我创建这个示例是为了更好地解释问题（我有两个 table 超过 10000 行）我希望得到这个结果（我手动完成的）：

我试过了，但没有结果：

def get_specialities(data1, data2):

    for i in data1['speciality_id']:

        for l in data1['id_questions']:

            for k in range(len(data2['id_questions_similar'])):

                drug_speciality = []

                if k == l :

                    drug_speciality.append(data1['speciality_id'])
                    return drug_speciality

print(get_specialities(data_1, data_2))

Answer 1

我是这样做的。可能可以在 pandas 上做所有事情并且更优化..？数据 1['id_questions'] 正在寻找数据 2['id_questions_similar'][i] 中的事件。根据这些事件，计算索引并获得“问题['speciality_id']”中的值。

import pandas as pd

data1 = {'id_questions': [1, 2, 3],
        'speciality_id': [[77, 66], [111, 122, 133], [1245]]
        }

active_questions = pd.DataFrame(data1)

data2 = {'id_medicament': [1, 2, 3],
        'id_questions_similar': [[1, 2], [1, 3], [3]]
        }

questions = pd.DataFrame(data2)

even_numbers = list(range(len(data2['id_questions_similar'])))

for i in range (0, len(data2['id_questions_similar'])):
        qqq = list(filter(lambda aaa: aaa in data1['id_questions'], data2['id_questions_similar'][i]))
        print(qqq)
        index_ = []
        count_ = len(qqq)
        for k in range(0, count_):
            ind = data1['id_questions'].index(qqq[k])#index
            if count_ == 1:
                    even_numbers[i] = data1['speciality_id'][ind]
                    continue
            index_.append(data1['speciality_id'][ind])#values
        if count_ > 1:
                even_numbers[i] = index_

questions['speciality_id'] = even_numbers
print(questions)

下面我做了一个更优化的版本

import more_itertools as mit
import pandas as pd

data1 = {'id_questions': [1, 2, 3],
        'speciality_id': [[77, 66], [111, 122, 133], [1245]]
        }

active_questions = pd.DataFrame(data1)

data2 = {'id_medicament': [1, 2, 3],
        'id_questions_similar': [[1, 2], [1, 3], [3]]
        }

questions = pd.DataFrame(data2)

iterable = active_questions['id_questions']
speciality = []
for i in questions['id_questions_similar']:
    index = list(mit.locate(iterable, lambda x: x in i))
    if len(index) > 1:
        speciality.append(list(active_questions.iloc[index, 1]))
    else:
        speciality.append(active_questions.iloc[index[0], 1])


questions['speciality_id'] = speciality
print(questions)

根据条件从两个表创建一个新列

Create a new column from two tables based on a condition

python

loops

nested-lists

pandas