Development environment:

win10 x64

cuda10.0

anaconda4.0.0

catalog

1. PIP error: cannot open D: anaconda3/scripts/pip- script.py

2. PIP switches domestic sources

3. Solve SciPy error: importerror: cannot import name ‘show_ config’ from ‘numpy’ (unknown location） 

4. Solve the problem of error: command error out with exit status 1

5. After installing scikit learn, it can be imported, but the running error is: importerror: DLL load failed: the specified program cannot be found

—————————————————————————————————————————————————————-

This error is due to the dependency package version is too high (too new).

Phenomenon restore: install scikit learn with anaconda, and then install Matplotlib to report the above error.

Solution: the installed version of Matplotlib is too high, and it can be solved by reducing the version.

However, this is not the end. Generally, scikit learn installed with the CONDA command may import normally, but it is easy to report an error when using it

The solution is to uninstall sklearn, and then use pip to install it again. However, if you directly use the PIP install scikit learn command to install it, there may be errors, and the process of installing again will be very complicated

See:

After installing sklearn, it can be imported normally, but an error is reported

Welcome to my personal blog: the road of machine learning