[PYTHON] Rebuilding an environment for machine learning with Miniconda (Windows version)

Introduction

Since I started using it for machine learning about a year ago, I've been installing it without knowing anything, but it's so chaotic that I'll rebuild it here.

Assumed reader

--Python can be used to some extent --Mainly use machine learning and data science ――I want to use multiple environments properly

What is Miniconda

Miniconda is a free minimal installer for conda. It is a small, bootstrap version of Anaconda that includes only conda, Python, the packages they depend on, and a small number of other useful packages, including pip, zlib and a few others. Use the conda install command to install 720+ additional conda packages from the Anaconda repository. (From Miniconda)

A lightweight version of Anaconda. It's lighter than Anaconda, which has a large number of packages installed from the beginning, and requires less storage occupancy. __ Anaconda is recommended for those who find it difficult to install packages one by one. __

(* Note) About package management tools other than Anaconda

There are other Python package management tools (Pipenv / venv, etc.), Here, we will introduce only the advantages and disadvantages of Anaconda (Miniconda).

__Anaconda Benefits __ Strong in machine learning and data science libraries (easy to install, high performance) You can easily create a virtual environment

__Anaconda Disadvantages __ There are few types of packages Some os functions may not be available

reference https://qiita.com/KRiver1/items/c1788e616b77a9bad4dd

Now, environment construction

Clear the original environment

Environment before rebuilding

--Python3.7 (Download from Python.org)

• Anaconda

I didn't keep a record of the deletion, so I will post a link that I used as a reference.

Uninstall Python (for Python.org)

https://gammasoft.jp/support/uninstall-python-completely-on-windows/

Uninstall Anaconda

https://insilico-notebook.com/conda-uninstall/ (* Note) I cleared it by the method of (2) above, but for some reason Anaconda itself could not be uninstalled, so I tried again by the method of (1).

Install Miniconda

-[1] Miniconda Download the installer

-[2] Run the installer

(Note 1) Basically, I think it is better to install with Just Me. (To keep the system environment clean)

(Note 2) If you check Add Miniconda3 to my PATH environment variable, the PATH of the related application will be added. Check it if you don't want to set environment variables later. (Officially deprecated.)

-[3] Check if the PATH is correct

Success if the conda command is recognized

Miniconda runtime environment management policy

Click here for virtual environment https://qiita.com/yakisobamilk/items/867dce8e53824146ce05

--__ base environment is not used __

It gets messy when the environment is corrupted.

--__ Prepare a virtual environment for default instead __

For the time being, prepare a virtual environment for the default in case you want to run Python, and use it instead of the base environment.

--_ Create a virtual environment for each project __

This area is the same.

__ Basically, we will operate in a virtual environment, and if the environment is damaged, we will maintain it with a policy of deleting the virtual environment. __

Finally

There are pros and cons to using Anaconda, but as a result of using it in various ways, I feel that it suits me best. Have a good Miniconda life! !! : snake: