can't install specific old version of sklearn to解决pyadio分析库中不兼容的“SVC属性错误”

Posted 2023-02-18

【中文标题】can\'t install specific old version of sklearn to解决pyadio分析库中不兼容的“SVC属性错误”

【英文标题】：can't install specific older version of sklearn to solve the incompatible "SVC attribute error" in pyadio analysis libcan't install specific old version of sklearn to解决pyadio分析库中不兼容的“SVC属性错误”

【发布时间】：2020-05-01 20:17:04

【问题描述】：

以下代码用于音频文件分割。

from pyAudioAnalysis import audiosegmentation as aS
[flagsInd, classesAll, acc, CM] = aS.mtFileClassification("diarizationExample.wav", "svmSM", "svm", True, 'dar.segments.txt')

它给了我这个警告：

C:\Users\Kenzhegaliyev_EK\AppData\Local\Continuum\anaconda3\lib\site-packages\pydub\utils.py:165: RuntimeWarning: Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work
  warn("Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work", RuntimeWarning)
C:\Users\Kenzhegaliyev_EK\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\utils\deprecation.py:144: FutureWarning: The sklearn.svm.classes module is  deprecated in version 0.22 and will be removed in version 0.24. The corresponding classes / functions should instead be imported from sklearn.svm. Anything that cannot be imported from sklearn.svm is now part of the private API.
  warnings.warn(message, FutureWarning)
C:\Users\Kenzhegaliyev_EK\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\base.py:318: UserWarning: Trying to unpickle estimator SVC from version 0.19.1 when using version 0.22.1. This might lead to breaking code or invalid results. Use at your own risk.
  UserWarning)

这个错误：

C:\Users\Kenzhegaliyev_EK\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\svm\_base.py in predict(self, X)
    583         """
    584         check_is_fitted(self)
--> 585         if self.break_ties and self.decision_function_shape == 'ovo':
    586             raise ValueError("break_ties must be False when "
    587                              "decision_function_shape is 'ovo'")

AttributeError: 'SVC' object has no attribute 'break_ties'

显然，这是 sklearn 中新旧版本的 SVC 之间的冲突。

在警告中它告诉我应该使用 sklearn version = 0.19.1

我尝试安装当前的并升级到 older sklearn。我不能使用shell，它受到限制，所以我在jupyter中运行所有代码：

!pip install sklearn
Installing collected packages: sklearn
Successfully installed sklearn-0.0

import sklearn
sklearn.__version__
'0.22.1'

!pip install --upgrade sklearn==0.19.1
ERROR: Could not find a version that satisfies the requirement sklearn==0.19.1 (from versions: 0.0)
ERROR: No matching distribution found for sklearn==0.19.1

官方sklearn website上没有0.19.1版本：

以下版本提供了基于 Web 的文档：

Scikit-learn 0.23.dev0 (dev) 文档 (PDF 48.5 MB)

Scikit-learn 0.22.1（稳定版）文档（PDF 48.5 MB）

Scikit-learn 0.21.3 文档 (PDF 46.7 MB)

Scikit-learn 0.20.4 文档 (PDF 45.2 MB)

Scikit-learn 0.19.2 文档 (PDF 42.2 MB)

Scikit-learn 0.18.2 文档 (PDF 46.5 MB)

Scikit-learn 0.17.1 文档 (PDF 46.0 MB)

Scikit-learn 0.16.1 文档 (PDF 56.8 MB)

我尝试安装任何其他旧的 sklearn，有什么可用的：

!pip install --upgrade sklearn==0.18.2
ERROR: Could not find a version that satisfies the requirement sklearn==0.18.2 (from versions: 0.0)
ERROR: No matching distribution found for sklearn==0.18.2

!pip install --upgrade sklearn==0.19.2
ERROR: Could not find a version that satisfies the requirement sklearn==0.19.2 (from versions: 0.0)
ERROR: No matching distribution found for sklearn==0.19.2

如何安装正确的旧 sklearn 以使用旧的兼容 SVC 来消除错误？

【参考方案1】：

sklearn 的版本很少。这是因为the correct name 是scikit-learn：https://pypi.org/project/scikit-learn/#history

所以跑

!pip install --upgrade scikit-learn==0.19.1

【讨论】：

谢谢你，但为什么我不能导入 scikit-learn？我做“导入sklearn”

我的意思是如果它不能导入scikit-learn，为什么它可以升级，降级名称“scikiet-learn”？