Pipe Breaks

TODO

• Need to identify the trained models, their purpose, and any target dependencies.

Introduction

• All the setup details for individual repositories can be found the respective README. Additionally, refer to the docs here

• The list of existing environments on machines in which the repositories can be run are here.

• Details on how to setup the environments are also documented and is available in https://general.gqc.com.

  Index	Repository	Description	Machine	Environment
  1	pipe-breaks	Original pipe breaks project	HP Laptop	pipe_breaks (Py=3.8.10, scikit-learn=1.2.2)
  		MSI	pipe-breaks (Py=3.9.16)
  2	loe-coe-app_deven-fork	Likelihood-of-Event app fork	HP Laptop	pipe_breaks (Py=3.8.10, scikit-learn=1.2.2)
  		MSI	pipe-breaks (Py=3.9.16)
  3	pipe-breaks-transfer-learning	Using models trained on one utility on another	HP Laptop	pipe_breaks_transfer_learning (Py=3.8.10, scikit-learn=0.24.2)
  		MSI	pipe-breaks (Py=3.9.16), pipe-breaks-transfer-learning (Py=3.8.10)

Documentation

• Sphinx documentation for pipe-breaks can be found in loe-coe-app here:
  • gh-pages branch
  • Static app - link available here
• There are also documents available in pipe-breaks that can be helpful.

info

• Commands are same irrespective of machine. Details can also be found in the respective repository's README.
• Make sure to check all requirements versions and the supported python versions for those libraries. Sometimes errors on streamlit are not descriptive enough to pinpoint the issue and may consume good amount of time to find and resolve them.

  Eg: For pipe-breaks-transfer-learning, having this in requirements.txt will result in errors if we run it on, say python 3.11.4, but not on 3.8.10 (due to library versions not being supported).

      scikit-learn==0.24.2
      -- Error: ModuleNotFoundError: No module named 'sklearn.neighbors._dist_metrics'

  • You can resolve this by setting the python version in that environment to 3.8.10.