统计Pandas列中含字典列表的单元格内字典元素数量

Got it, let's break down how to count the number of dictionary elements in each cell of your df['likes'] column. The column has a mix of NaNs, and cells containing dictionaries with a data key that holds a list of dictionaries—we need to grab the length of that data list for each valid cell.

方法一：自定义函数处理（清晰直观） #

First, we can write a small helper function to handle each cell, then apply it to the column:

import pandas as pd

def count_like_entries(cell):
    # 先处理空值
    if pd.isna(cell):
        return 0  # 如果你想保留空值标记，可以换成 pd.NA
    # 检查单元格是否是包含'data'键的有效字典
    if isinstance(cell, dict) and 'data' in cell:
        return len(cell['data'])
    # 针对不符合预期格式的情况（非字典或缺少'data'键）返回默认值
    return 0

# 生成统计列
df['like_entry_count'] = df['likes'].apply(count_like_entries)

方法二：简洁的Lambda表达式 #

If you prefer a more compact approach, you can combine str.get() with a lambda to skip writing a separate function:

# 先提取'data'列表，再统计长度（同时处理空值）
df['like_entry_count'] = df['likes'].str.get('data').apply(lambda x: len(x) if not pd.isna(x) else 0)

额外处理：如果字典是字符串格式 #

In case some cells have dictionary-like strings (instead of actual Python dictionaries), you'll need to convert them first using ast.literal_eval:

import ast

# 将字符串格式的字典转换为真实的Python字典对象
df['likes'] = df['likes'].apply(lambda x: ast.literal_eval(x) if isinstance(x, str) else x)

# 然后运行上面的统计代码
df['like_entry_count'] = df['likes'].str.get('data').apply(lambda x: len(x) if not pd.isna(x) else 0)

两种方法都会生成一个新列，里面是每个likes单元格内的字典元素数量——空值单元格会显示0（你也可以根据需求调整为显示pd.NA来保留空值标记）。

内容的提问来源于stack exchange，提问作者Edward