A Conda Coup D’etat over Pip

Why? My various projects, python versions, my macbook, my IDEs and virtual environments got unwieldy overtime.

Conda	Pip
conda is a system package manager. conda virtual environments isolate projects like Docker containers. Read more. conda has different package repositories.	pip is a Python package manager. separately need a virtual environment is needed for pip to function the same way. Read more. pip has one global package repository (pypi).

table 1: basic differences between conda and pip

Anaconda	Miniconda
CLI package manager, conda The latest version of Python supported and dependencies Anaconda navigator the GUI for apps, packages, environments and channels Packages for DS and ML	Conda Dependencies (Python) Navigator No packages, so it’s 10x smaller than Anaconda

Table 2: differences between Anaconda and Miniconda

Getting started (post install):

You start in the base environment. Never use the base environment for dev. conda env list
You create a new environments for projects. conda create --name example
You can specify different python versions. conda create --name example python=3.9
You can activate the environment for the project. conda activate example
List packages in your environment. conda list
Install packages. conda install jupyterlab pandas dask pyplot ipykernel
Install packages from other channels (repos). conda install -c conda-forge conda-stats
Match the kernel to environment name. python -m ipykernel install --user --name=example
Launch JupyterLab. jupyter-lab
Close JupyterLab. (Ctrl+C)
Deactivate current environment before switching. conda deactivate

ML/AI	ETL &Dashboard	R
pandas pytorch nltk scikit-learn	dask panel airflow pandas	r ggplot2 dplyr

table 3: examples of creating different environments for different project purposes

Supported IDEs that I also use: