如何在Pandas中将DataFrame单列值转换为多列值

原始数据结构 #

目标数据结构 #

实现方法一：分组自定义转换 #

核心逻辑是按id和date分组，根据每组内的decision集合判断目标列取值：

• CR优先取revise，无revise则取checked
• Principal取improve或complete，无对应值则设为NA

import pandas as pd

# 构造原始DataFrame（已有数据可跳过此步）
df = pd.DataFrame({
    'id': [1,1,2,3,4,4],
    'date': ['2022-11-10','2022-11-10','2021-09-12','2020-08-22','2019-11-10','2019-11-10'],
    'decision': ['improve','checked','checked','checked','complete','revise']
})

def process_group(group):
    decisions = set(group['decision'])
    cr_val = 'revise' if 'revise' in decisions else 'checked'
    principal_val = next((d for d in ['improve', 'complete'] if d in decisions), pd.NA)
    return pd.Series({'CR': cr_val, 'Principal': principal_val})

# 分组处理并重置索引
result_df = df.groupby(['id', 'date']).apply(process_group).reset_index()
print(result_df)

实现方法二：透视表映射转换 #

先给每个decision标记对应的目标列，再通过透视表重组结构，最后调整冲突取值：

import pandas as pd

df = pd.DataFrame({
    'id': [1,1,2,3,4,4],
    'date': ['2022-11-10','2022-11-10','2021-09-12','2020-08-22','2019-11-10','2019-11-10'],
    'decision': ['improve','checked','checked','checked','complete','revise']
})

# 标记每个decision对应的目标列
df['target_col'] = df['decision'].map({
    'checked': 'CR',
    'revise': 'CR',
    'improve': 'Principal',
    'complete': 'Principal'
})

# 透视表重组结构
result_df = df.pivot(index=['id', 'date'], columns='target_col', values='decision').reset_index()

# 处理CR列：同时存在checked和revise时保留revise
result_df['CR'] = result_df['CR'].apply(lambda x: 'revise' if isinstance(x, str) and 'revise' in x else x)

# 填充缺失值为NA
result_df = result_df.fillna(pd.NA)

print(result_df)

内容的提问来源于stack exchange，提问作者Swetha