Calculate variance explained #88

ejiawustl · 2024-06-07T21:29:04Z

Added new file util.py in the probability models folder to include the function to calculate the explained variance using sklearn. Also added a new test suite to check that the function runs properly. Also have been playing around with the visualizing_and_testing_data_generation_methods notebook - will need to polish it further but it can become a good tutorial in the future

This is already in the dev dependencies. I forgot to go over that. To add 'production' depdencies with python, you add to the default dependencies section with just: ``` poetry add <package> ``` You can also add dependencies to a group, eg: ``` poetry add --group dev <package> ``` See https://python-poetry.org/docs/cli/#options-4 That way, you can control what dependencies get installed. For a typical user, I don't think we'll want to install jupyter in the environment. They should have jupyter in their environment, and then install yeastdnnexplorer into it.

cmatKhan

I think the variance explained is still in the notebook -- is that correct?

I made a mistake when I had you change all the single/noise -- I should have used that as the example of the fork/branch/pull process. So many changes makes it difficult to focus on what is meaningful. My fault.

yeastdnnexplorer/probability_models/generate_data.py

cmatKhan · 2024-07-06T16:15:24Z

@ejiawustl see my last comment.

If you make a push to this branch on your fork, it will update this pull request automatically

cmatKhan · 2024-07-06T16:15:53Z

closes #84

…ance_explained function and test suite

…awustl/yeastdnnexplorer into calculate_variance_explained merge with remote

cmatKhan · 2024-07-23T14:05:20Z

@ejiawustl -- it is not always obvious when it is necessary to add a function that wraps a scikit/numpy/whatever function for ease of use in the package context, and when not to. That is part of the art.

In this case, my decision is not to wrap the scikit variance explained function. That function seems to link together a couple steps -- basically evaluating the model -- and then uses the output to call the variance explained function. Since the modelling is something that the user should be doing themselves -- they'll need to make decisions, we can't automate that -- they should be able to also use the scikit variance explained function. We can certainly demonstrate how -- that should go into a tutorial.

Think about this. If there is a good reason to have this function in the codebase, please explain.

ejiawustl · 2024-07-23T15:54:46Z

@ejiawustl -- it is not always obvious when it is necessary to add a function that wraps a scikit/numpy/whatever function for ease of use in the package context, and when not to. That is part of the art.

In this case, my decision is not to wrap the scikit variance explained function. That function seems to link together a couple steps -- basically evaluating the model -- and then uses the output to call the variance explained function. Since the modelling is something that the user should be doing themselves -- they'll need to make decisions, we can't automate that -- they should be able to also use the scikit variance explained function. We can certainly demonstrate how -- that should go into a tutorial.

Think about this. If there is a good reason to have this function in the codebase, please explain.

Ok, after thinking about it some more I agree with this as well. We don't need the wrapper function, I think it was only something I was using at the time and was told to make it a function in the codebase.

…lained and adding the function to the visualizing_and_testing_data_generation_methods notebook

review-notebook-app · 2024-07-23T17:22:36Z

Check out this pull request on

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

cmatKhan · 2024-07-23T17:52:10Z