anaconda

see also:

• artificial intelligence
• AI machine learning
• Python language
• AI deep learning
• Anaconda website
  • https://docs.anaconda.com/free/anaconda/reference/troubleshooting/
• http://www.pytorch.org

• basic level Anaconda is a free application which is used for:
  • installing Python language, modules, and tools
  • creating Python sand boxed environments
  • launching Python tools such as Jupyter notebooks
• it is a Python/R data science distribution containing conda.
• Conda is an environment and package manager that helps users manage the other 7,500+ open-source packages made available to them through the distribution.
• Anaconda Navigator
  • is a GUI app for managing environments, packages and launching apps
• Anaconda terminal command prompt method of accessing conda and pip:
  • this can:
    • activate your environment
    • install Python modules
• It uses “channels” to specify URLs which point to package repositories
  • the default channel points to several channels at the repo.anaconda.com

• see Conda cheat sheets
• see conda command line arguments for install

conda info
conda update -n base conda #optionally run this to update the conda install to latest install
conda update anaconda #optionally run this to update the anaconda install to latest install

conda activate environment_name

• see Anaconda libraries and packages

conda install -c main pandas
conda install numPy
conda install matplotlib
conda install -c main scikit-learn
conda install scikit-learn-intelex # this is an accelerator for scikit-learn for win64
conda install -c main tensorflow #in July 2023, this will not install on python v3.11 as only compatible with python 3.5.* to 3.10.*
conda install -c main nltk #natural language toolkit
conda install -c pytorch pytorch-cuda #NB. -c is the channel parameter so here it is using the pytorch channel to search and install the latest pytorch-cuda package
#can use -n ENVNAME to choose the environment to install if you have not activated that environment

• unfortunately on my system, I was not able to install PyTorch cuda version with this as I continually got errors solving environment and had to resort to a pip install

• see https://docs.anaconda.com/free/navigator/
• create a new environment
  • click on left menu item Environments then at bottom of screen click on Create then name your new environment
  • you can then use Command Prompt to install additional modules into this such as MatPlotlib. NumPy, PyTorth cuda version, etc
• use your new environment for launching apps
  • manually select your new environment via clicking on the environment under the menu item Environments, or,
  • set that environment top be the default File:Preferences then set Default Conda Environment
• run app to use your environment
  • go to Home menu item option
  • ensure your environment is the one selected then launch your app (if no launch button visible, you need to install the app first using the install button)
  • eg. launch Jupyter Notebook
    • this will open Jupyter Notebook in a web browser and then open the main directory url http://localhost:8888/tree so you can choose to open an existing notebook (.ipynb files), or create a new one using the Python 3 kernel