Pandas数据分类结果异常：收入未达标却被标记为Y问题排查

你的代码出现不符合预期结果的核心原因是运算符优先级问题：Python中&（按位与）的优先级高于>、==这类比较运算符，导致条件表达式没有按照你预期的逻辑执行。

比如你写的row['total_income']>28000 & row['children']==0，会被Python解析为：
row['total_income'] > (28000 & row['children'])

以你提到的第1行数据为例：

• row['children']的值是1，28000 & 1的按位与结果为0
• row['total_income']为17932.802，显然大于0，因此这个条件判断结果为True，触发了test='Y'的赋值，完全偏离了你原本的逻辑。

解决方案 #

需要给每个比较表达式单独加括号，明确逻辑判断的顺序，你可以选择两种写法：

写法1：用&并添加括号

def assign_status(row):
    test='Negado'

    # Desempregado
    if row['income_type']=='unemployed':
        test='N'
        
    # Já devedor
    elif row['debt']==1:
        test='N'
    # Relação de renda e filhos.
    elif (row['total_income']>28000) & (row['children']==0):
        test='Y'
    # Teste recuperado
    elif (row['total_income']>31000) & (row['children']==1):
        test='Y'
    # Teste desconhecido
    elif (row['total_income']>34000) & (row['children']==2):
        test='Y'
    elif (row['total_income']>37000) & (row['children']==3):
        test='Y'
    elif (row['total_income']>40000) & (row['children']==4):
        test='Y'
    elif (row['total_income']>43000) & (row['children']==5):
        test='Y'
    else:
        test='N'
    return test

写法2：用and（更符合Python逻辑判断的常规写法）

def assign_status(row):
    test='Negado'

    # Desempregado
    if row['income_type']=='unemployed':
        test='N'
        
    # Já devedor
    elif row['debt']==1:
        test='N'
    # Relação de renda e filhos.
    elif row['total_income']>28000 and row['children']==0:
        test='Y'
    # Teste recuperado
    elif row['total_income']>31000 and row['children']==1:
        test='Y'
    # Teste desconhecido
    elif row['total_income']>34000 and row['children']==2:
        test='Y'
    elif row['total_income']>37000 and row['children']==3:
        test='Y'
    elif row['total_income']>40000 and row['children']==4:
        test='Y'
    elif row['total_income']>43000 and row['children']==5:
        test='Y'
    else:
        test='N'
    return test

两种写法都能解决你的问题，推荐使用and，它更直观，符合Python逻辑判断的常规用法。

内容的提问来源于stack exchange，提问作者Riuk2252