Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

838 create tabular tab to dashboard and redesign loaded data results #819 #844

Merged
merged 98 commits into from
Sep 18, 2024

Conversation

laurasootes
Copy link
Contributor

@laurasootes laurasootes commented Aug 22, 2024

This PR adds a tabular page to the dashboard (#838 ) and finalises all dashboard redesigns (#819 ).

Tabular page:
Page is implemented similar to the other pages (text, images, time series).

  • Other than the other pages, the tabular page requires the user to upload the training data, which has to be provided as npy file.
  • For this page it is not obligatory to upload a label text file, because you can also use it for regression models, in which case labels are not necessary. If labels are not uploaded, the page will treat the example as a regression model.
  • The available methods are currently RISE and LIME.
  • For the usage, a user uploads their tabular data csv file (full file). The data is read as pandas dataframe and the full table is shown in the dashboard. The user then clicks on a row in the table, which is the data for which the prediction and explanation is done.
  • A simple test to check the basic functionality of the page is added to the dashboard tests.
  • Note that a tabular example will be added in Add tabular example in dashboard #789

Dashboard redesign changes:

  • _get_method_params is converted into get_params, such that method parameter seleciton is per method
  • For all pages a streamlit container (basically a box) is created at the top which includes the predicted class, number of top classes to show, method selection and method parameters. The different components are only shown when relevant.
  • function _get_top_indices_and_labels is adjusted such that for a regression model the predicted value is printed in the dashboard instead of the predicted class

Other changes:

  • naming of keys for streamlit elements is done in a prettier way

fixes #838 and fixes #819 and fixes #523

@laurasootes laurasootes linked an issue Aug 22, 2024 that may be closed by this pull request
@elboyran elboyran mentioned this pull request Aug 23, 2024
@@ -63,6 +63,7 @@ def __init__(
LimeTabularExplainer, kwargs)

# temporary solution for setting num_features and top_labels
# when fixed, also fix in dashboard Tabular.py -> _feature_names
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it fixed in the next PR?

Copy link
Contributor Author

@laurasootes laurasootes Sep 18, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No, the temporary fix itself was already here, it has been added in december PR 661. I did not try to fix the general problem

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I believe issue 801 mentions the general issue

@laurasootes laurasootes merged commit 45fd4ff into main Sep 18, 2024
15 of 17 checks passed
@laurasootes laurasootes deleted the 838-create-tabular-tab-to-dashboard branch September 18, 2024 08:26
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
2 participants