Conda 安装与 pip 安装（导入 seaborn 包时出现问题）

Question

I am facing an issue in importing module in jupyter notebook. The issue is that I access the jupyter notebook through anaconda software and when i tired installing seaborn, I used the command "pip install seaborn" in cmd and the package was installed(verifed through "pip list") but when i tried importing the seborn package in notebook it gave the error ModuleNotFoundError: No module named 'seaborn' I even restarted the kernel,notebook and anacond but still the issue persisted. But when i did "conda list" in the notebook the package was not there but then after "conda install seaborn" the package got installed and the package got imported. I wanted to know why this error occured as i didnt face this situation while importing numpy or pandas(for numpy or pandas only pip install was enough)

python version:3.10.5

conda version:4.14.0

jupyter version:4.10.0

Answer 1

Caveat: Without more information (such as OP's command history and environment settings), one can't answer this question with 100% confidence, but there is a most likely explanation:

Running pip install seaborn via cmd installed seaborn in a directory that is not part of the Python path of the conda environment that the Jupyter Notebook was running in. It seems most likely you were not using a conda command prompt to pip install seaborn , but you were using a conda command prompt to pip install numpy and pandas , which is why they were installed in a directory included in the Jupyter Notebook's conda environment's Python path.

For example, a standard conda package install location in a Windows-style filesystem for a conda environment named <env-name> is the site-packages directory for that environment, C:\Users\<User>\<Conda_Distribution>\envs\<env-name>\Lib\site-packages . This directory is in the Python path of a its corresponding environment by default.

This knowledge of where to find your actual conda packages in a directory comes with a warning: Unless you have expert knowledge of the Python import system and conda's package management infrastructure, it is a generally a bad idea to manually add or delete directories in your conda environment's site-packages directory.

A related warning: Although the interoperability of pip with conda has been greatly improved in the last few years, it is generally a bad idea to use pip to install standard Python packages in your conda environment. conda install exists, and you should use it for a reason .

For more detailed information about what your Python path is and how it's set, see this informational guide from Real Python.