使用pandas_profiling/ydata_profiling生成报告遇PydanticImportError

问题详情 #

尝试用两种方式生成Pandas数据集的分析报告：
第一种实现代码：

import pandas_profiling 
profile_report = df.profile_report()

第二种实现代码：

from ydata_profiling import ProfileReport

两种方式均触发PydanticImportError，核心错误提示为BaseSettings已迁移至pydantic-settings包，完整报错栈如下：

PydanticImportError: BaseSettings has been moved to the pydantic-settings package.
Traceback:
File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 541, in run_script
exec(code, module.dict)
File "C:\Users\HP\OneDrive\Desktop\AutoML\app.py", line 5, in
import pandas_profiling
File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\site-packages\pandas_profiling_init.py", line 8, in
from ydata_profiling.compare_reports import compare
File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\site-packages\ydata_profiling_init_.py", line 7, in
from ydata_profiling.compare_reports import compare
File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\site-packages\ydata_profiling\compare_reports.py", line 9, in
from ydata_profiling.config import Correlation, Settings
File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\site-packages\ydata_profiling\config.py", line 7, in
from pydantic import BaseModel, BaseSettings, Field, PrivateAttr
File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\site-packages\pydantic_init_.py", line 218, in getattr
return _getattr_migration(attr_name)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\HP\AppData\Local\Programs\Python\Python311\Lib\site-packages\pydantic_migration.py", line 294, in wrapper
raise PydanticImportError(

报错原因 #

Pydantic 2.x版本把BaseSettings从主包移到了独立的pydantic-settings包，但你当前安装的ydata-profiling（pandas-profiling是它的旧名称）版本还在从pydantic主包导入BaseSettings，版本不兼容导致报错。

解决方法 #

方法1：安装兼容的Pydantic 1.x版本 #

如果不想升级ydata-profiling，可以回退到Pydantic 1.x版本，这个版本的BaseSettings仍在主包内：

pip install pydantic==1.10.12

方法2：升级ydata-profiling到适配Pydantic 2.x的版本 #

安装最新版的ydata-profiling，新版已经修复了这个导入问题：

pip install --upgrade ydata-profiling

方法3：临时兼容方案（不推荐长期使用） #

如果上述两种方法暂时无法执行，可以先安装pydantic-settings包，再在代码开头添加兼容补丁：
首先安装包：

pip install pydantic-settings

然后在代码最开头添加：

try:
    from pydantic import BaseSettings
except ImportError:
    from pydantic_settings import BaseSettings
    import pydantic
    pydantic.BaseSettings = BaseSettings

内容的提问来源于stack exchange，提问作者Unreal Capacity