pandas构建DataFrame时列有元素却显示为NaN的问题排查
构造DataFrame时自定义列名后整列显示NaN的问题
问题复现代码
import pandas as pd customers = { "CustomerId": [1,2,3,4], "FirstName" : ["Ahmet","Ali","Hasan","Canan"], "LastName" : ["Yılmaz","Korkmaz","Çelik","Toprak"] } orders = { "OrderId" : [10,11,12,13], "CustomerId" : [1,2,5,7], "OrderDate" : ["2010-07-04","2010-08-04","2010-07-07","2012-07-04"] } df_customers = pd.DataFrame(customers, columns = ["CustomerId","FirstName","LastName"]) df_orders = pd.DataFrame(orders, columns = ["orderId","CustomerId","OrderDate"]) print(df_orders)
核心原因
pandas在根据传入的字典数据生成DataFrame时,对columns参数的匹配规则是严格区分大小写的精确匹配:
- 原始orders字典存储订单ID字段时,键名为
OrderId(大写O开头) - 构造df_orders时传入的columns参数中,对应列名写为
orderId(小写o开头)
由于两个字符串完全不匹配,pandas无法从原始字典中读取到OrderId对应的数值,就会按照“自定义列不存在于原数据源”的逻辑,把这个列的所有值填充为NaN。其余列名和字典键完全一致的字段,可正常读取到原始数据。
实际运行输出
orderId CustomerId OrderDate 0 NaN 1.0 2010-07-04 1 NaN 2.0 2010-08-04 2 NaN 5.0 2010-07-07 3 NaN 7.0 2010-07-04
修复方法
- 对齐原始字典键名与columns参数的列名,保证字符串完全一致即可解决问题:
# 修正columns参数中的列名大小写,和字典键匹配 df_orders = pd.DataFrame(orders, columns = ["OrderId","CustomerId","OrderDate"]) - 如果业务场景下需要保留小写开头的orderId列名,可优先基于原始字典构造DataFrame,再做列名重命名,避免直接传入不匹配的列名:
df_orders = pd.DataFrame(orders) df_orders = df_orders.rename(columns={"OrderId": "orderId"})
内容的提问来源于stack exchange,提问作者Taha Pasha
相关产品推荐
相关产品推荐

