Setting up conda

You need to use the pre-configured conda channels, which means that a environment.yaml file with the following will not work.

name: los_predictor_v1
channels:
  - defaults  # <--- default channel not available
dependencies:
  - python=3.10.10
  - pandas
    pip:
    - matplotlib

So comment out the line channels section!

Open a terminal window

conda env create -f environment.yaml

0.1 Notes

  • Conda has been configured to use a local mirror of conda and conda-forge in a framework called Nexus. This means that you will have access to most, but not all, packages and some of them will be outdated.
  • There are some conda environments that have already been made with pytorch/tensorflow/automl. Ideally those should be used in the first instance and packages can be added to that.
  • If you wish to create your own environment, ensure that you have activated your base conda environment using conda activate prior to creating any other conda environments
  • If you do wish to install other conda-like installers (e.g. mamba), ensure that the base conda environment has been activated so that the configuration files for Nexus has migrated into the mamba configuration.
    • You’ll probably have more success with conda-forge as your default channel as Nexus mirrors that more successfully. To do this, run: conda config --add channels conda-forge
  • On first use, jump into a terminal and conda init