使用 pandas 从多索引系列数据框中找出重复的字符串元素

Question

在 Pandas 数据框中，我拥有的一列是系列数据类型，即 food_column，我必须从该列中提取输出列

   Input : food_column
 [ 'bread','bread','bread'] ,
 [ 'meat','butter','butter'] ,
 [ 'meat', 'butter','bread','meat']
        ['butter']
 ['bread','meat','bread','meat']

Output : main_column
       ['bread']
       ['butter']
        ['meat']
       ['butter']
       ['bread']

条件：

1. 如果任何字符串元素重复多次，则应将其选为输出元素，
2. 如果任何两个或三个元素计数相同，则应从这两个或三个元素中选择 np.random.choice
3. 如果任何行中只有一个元素，则将该元素分配/映射到输出列
4. 否则将其标记为“未知”以输出列

Answer 1

import pandas as pd
import random
from collections import Counter
import numpy as np

food_list = [[ 'bread','bread','bread'] ,
 ['meat','butter','butter'] ,
 ['meat', 'butter','bread','meat'],
 ['butter'],
 ['bread','meat','bread','meat'],
 ['']]

food_series = pd.Series(food_list)

df = pd.DataFrame({'food_column': food_series})

# randomize list item order, since dict item order is constant in Python 3.6+ 
df['random_food_list'] = [random.sample(z, len(z)) for z in df['food_column'].to_list()]
# get counts
df['food_counts'] = df['random_food_list'].apply(lambda x: Counter(x))
# get key with max value
df['main_column'] = df['food_counts'].apply(lambda x:  max(x, key=x.get))
# replace empty strings with 'unknown'
df['main_column'] = np.where(df['main_column'] == '', 'unknown', df['main_column'])