如何解决导入numba与numpy时出现的SystemError: initialization of _internal failed错误？

问题背景 #

基于时间序列历史数据开发回归分析预测器，项目包含my_project.py、utilities.py、plotter.py和constants.py四个文件，运行时触发Numba相关初始化错误。

关键导入代码 #

my_project.py

from time import perf_counter
from constants import (output_dir, DATAPATH, output_file)
from utilities import (dataframe_in_nutshell, excel_reader, info_printer, sys, module_creator, process_discovery, data_explanatory_analysis, excel_reader, df_cleaner, feature_extractor, ml_modelling)
from plotter import Plotter

utilities.py

import os
import sys
import inspect
from pathlib import Path
from typing import (Iterable, List, Tuple, Optional)
from itertools import zip_longest
import matplotlib.pyplot as plt
import statsmodels.tsa.api as smt
import statsmodels.api as sm
import pandas as pd
from sklearn.metrics import mean_absolute_error
from sklearn.preprocessing import scale
from pycaret.regression import (setup, compare_models, predict_model, plot_model, finalize_model, load_model)
import csv
from constants import (np, Path, nan_value, plots_dir, HOURS_PER_WEEK, LAGS_STEP_NUM, rc_params, NA_VALUES, COLUMNS_NAMES, string_columns, LAGS_LABELS, numeric_columns, output_dir, DATAPATH, dtype_dict, train_size)
from pprint import PrettyPrinter
pp = PrettyPrinter()
import seaborn as sns

sns.set()

错误栈信息 #

Traceback (most recent call last):
  File "c:\Users\username\OneDrive\Desktop\project\my_project.py", line 10, in <module>
    from utilities import (dataframe_in_nutshell, excel_reader, info_printer, sys, module_creator,
  File "c:\Users\username\OneDrive\Desktop\project\utilities.py", line 18, in <module>
    from pycaret.regression import (setup, compare_models, predict_model, plot_model, finalize_model,
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\pycaret\regression.py", line 10, in <module>        
    import pycaret.internal.tabular
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\pycaret\internal\tabular.py", line 48, in <module>  
    import pycaret.internal.preprocess
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\pycaret\internal\preprocess.py", line 27, in <module>
    from pyod.models.knn import KNN
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\pyod\__init__.py", line 4, in <module>
    from . import utils
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\pyod\utils\__init__.py", line 4, in <module>        
    from .stat_models import pairwise_distances_no_broadcast
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\pyod\utils\stat_models.py", line 11, in <module>    
    from numba import njit
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\numba\__init__.py", line 42, in <module>
    from numba.np.ufunc import (vectorize, guvectorize, threading_layer,
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\numba\np\ufunc\__init__.py", line 3, in <module>    
    from numba.np.ufunc.decorators import Vectorize, GUVectorize, vectorize, guvectorize
  File "C:\Users\username\anaconda3\envs\py310\lib\site-packages\numba\np\ufunc\decorators.py", line 3, in <module>  
    from numba.np.ufunc import _internal
SystemError: initialization of _internal failed without raising an exception

运行环境 #

• Windows 10系统
• VSCode编辑器
• Conda环境，Python 3.10，所有依赖来自conda-forge频道

解决方案 #

按优先级尝试以下修复步骤：

• 降级Numba版本：conda-forge部分Numba版本在Windows Python3.10环境存在兼容性问题，执行命令安装稳定版本：

  conda install -c conda-forge numba=0.56.4
  

• 重新安装Numba及依赖：先卸载现有包再重装，解决损坏的安装文件：

  conda remove numba
  conda install -c conda-forge numba
  

• 更新Conda环境及所有依赖：解决潜在依赖冲突，确保环境为最新状态：

  conda update -n base conda
  conda update --all
  

• 创建全新Conda环境：若上述方法无效，搭建干净环境重新安装依赖：

  conda create -n ts_prediction python=3.10 -c conda-forge
  conda activate ts_prediction
  conda install pandas numpy scikit-learn statsmodels pycaret seaborn matplotlib -c conda-forge
  

• 临时禁用PyOD的Numba依赖（可选）：若无需异常检测功能，在utilities.py最顶部添加代码禁用Numba：

  import os
  os.environ['PYOD_NUMBA_ENABLED'] = '0'
  

内容的提问来源于stack exchange，提问作者Rebel