Uploaded image for project: 'Moodle'
  1. Moodle
  2. MDL-64779

Allow users to choose whether to include the trained model weights in the export or not

    XMLWordPrintable

    Details

    • Testing Instructions:
      Hide

      Prerequisites

      1. Install https://github.com/dmonllao/moodle-local_testanalytics
      2. Log in as admin and go to site admin > analytics > analytics settings
      3. Disable onlycli setting

      Test

      1. Go to site admin > analytics > analytics models
      2. Execute Get predictions for the Binary model
      3. Click on Export for the Binary model
      4. You SHOULD see a modal window with a radio and the Export configuration option selected by default
      5. The checkbox and its text SHOULD disappear when you select Export training data
      6. Select Export training data and click on Export button
      7. A CSV file SHOULD be downloaded and the modal window should close
      8. Click on Export for the Binary model
      9. Select Export configuration, including the weights and click on Export
      10. A zip file SHOULD be downloaded, open it. An mlbackend directory SHOULD be part of the zip
      11. Click on Export for the Binary model
      12. Select Export configuration, untick the weights checkbox and click on Export
      13. A zip file SHOULD be downloaded, open it. No mlbackend directory SHOULD be listed.
      14. Set up a Students at risk of dropping out model. Basically ensure that the model has a time-splitting method and indicators
      15. Expand the Actions menu for the Students at risk of dropping out model
        • If you can see a Clear predictions option click on it.
      16. Click on Export for the Students at risk of dropping out model
      17. A zip file SHOULD be downloaded, open it. No mlbackend directory SHOULD be listed.
      Show
      Prerequisites Install https://github.com/dmonllao/moodle-local_testanalytics Log in as admin and go to site admin > analytics > analytics settings Disable onlycli setting Test Go to site admin > analytics > analytics models Execute Get predictions for the Binary model Click on Export for the Binary model You SHOULD see a modal window with a radio and the Export configuration option selected by default The checkbox and its text SHOULD disappear when you select Export training data Select Export training data and click on Export button A CSV file SHOULD be downloaded and the modal window should close Click on Export for the Binary model Select Export configuration , including the weights and click on Export A zip file SHOULD be downloaded, open it. An mlbackend directory SHOULD be part of the zip Click on Export for the Binary model Select Export configuration , untick the weights checkbox and click on Export A zip file SHOULD be downloaded, open it. No mlbackend directory SHOULD be listed. Set up a Students at risk of dropping out model. Basically ensure that the model has a time-splitting method and indicators Expand the Actions menu for the Students at risk of dropping out model If you can see a Clear predictions option click on it. Click on Export for the Students at risk of dropping out model A zip file SHOULD be downloaded, open it. No mlbackend directory SHOULD be listed.
    • Affected Branches:
      MOODLE_37_STABLE
    • Fixed Branches:
      MOODLE_37_STABLE
    • Pull from Repository:
    • Pull Master Branch:
      MDL-64779_master

      Description

      In MDL-60944 we add the possibility to export and import models' configurations. The exported config includes the target, the indicators, the time splitting method and the predictions processor that were used in the origin site. We include the trained model weights as well.

      Sander Bangma mentioned that he would prefer to add an option so users can choose whether to include the trained model's weights or not. At the moment these model-dependant actions are accessed through an action menu. For this case we have a "Export configuration" menu option, the exported file is downloaded once the option is clicked. If we want users to choose if the weights should be included or not we will need an extra JS modal or something like that before executing the action so they can select. It would be great to have UX team (Barbara Ramiro Ash Bettridge) input on this.

        Attachments

          Issue Links

            Activity

              People

              • Votes:
                0 Vote for this issue
                Watchers:
                4 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved:
                  Fix Release Date:
                  20/May/19

                  Time Tracking

                  Estimated:
                  Original Estimate - Not Specified
                  Not Specified
                  Remaining:
                  Remaining Estimate - 0 minutes
                  0m
                  Logged:
                  Time Spent - 50 minutes
                  50m