ImportError: DLLload failed when importing matplotlib.pyplot in Python 3.6.8 under Windows environment: The specified module could not be found

- The matplotlib dependency library version does not match the problem (to solve the problem see below)
- Ultimately, invincible universe is the only solution for the entire network

*In the first pip install matplotlib, actively installed to the latest version (matplotlib==3.3.2) at the time, and summarized the following solutions in StackOverflow and many domestic posts
Error in matplotlib 3.3.2 version
Error in matplotlib 2.2.2 version

The matplotlib dependency library version does not match the problem (to solve the problem see below)

I have experienced countless attempts to install and uninstall matplotlib-related dependent libraries such as numpy scipy, etc. I have
found the correct relationship of all matplotlib dependent libraries as shown in the figure below, so that sentient beings can avoid detours and leave more hair.
The matplotlib official website link is as shown below

Link: https://matplotlib.org/users/installing.html .

Picture: Insert picture description here
These dependent libraries are best checked, although we will not solve the problem after checking

The command to check the installation package (dependent library) under pip is as follows

pip list

Uninstall the installation package (dependent library) command is as follows

pip uninstall <依赖库==版本号>

installation

pip install <依赖库==版本号>

Ultimately, invincible universe is the only solution for the entire network

The reason for the error: all the dependent library versions in import matplotlib.pyplot as plt meet the requirements, but the reason for the error is that the online pip install numpy is due to the lack of DLL
in the online download of the numpy library . The MKL version is solved in the unofficial library connection of the University of California. this problem

Download the Numpy+MKL version and pay attention to the native Python version
(Note: the cp36 in the figure below is the python3.6.x version)
Note: The download speed is slow, if there are scientific Internet tools, it will be better
**

By Christoph Gohlke, Laboratory of Fluorescence Dynamics, University of California, Irvine
Unofficial Windows binaries for Python extension packages.

Insert picture description here
Link: https://www.lfd.uci.edu/~gohlke/pythonlibs/#numpy .

Offline installation:

pip install <下载好的文件路径+文件名>

Excited heart and shaking hands, the road to deep learning is hard to follow! ! ! Hahaha finally became
Insert picture description here