Uploaded image for project: 'Moodle'
  1. Moodle
  2. MDL-59661

Add a button to export model's training data

    XMLWordPrintable

Details

    • New Feature
    • Status: Closed
    • Major
    • Resolution: Fixed
    • 3.4
    • 3.4
    • Analytics
    • MOODLE_34_STABLE
    • MOODLE_34_STABLE
    • MDL-59661_master
    • Hide

      Here we are going to test that:

      • Multiple training task executions result in multiple training files generated.
      • These file contents can be combined into 1 and downloaded
      1. You need 1 course with a past start date and end date and with at least 1 student
      2. Students at risk of dropping out models should be enabled and should have a time splitting method defined
      3. Moodle will only use courses for training if they have a minimum of activity (more than 10 logs per student in a course) You can generate all these logs or just edit lib/classes/analytics/target/course_dropout.php is_valid_analysable and make it return true

      4. php admin/tool/task/cli/schedule_task.php --execute=\\tool_analytics\\task\\train_models

      5. Go to admin/tool/analytics/index.php and Actions > Export "Students at risk of dropping out" model
      6. Open the downloaded file and note the number of rows
      7. Enrol another 2 students to that course

      8. php admin/tool/task/cli/schedule_task.php --execute=\\tool_analytics\\task\\train_models

      9. Go to admin/tool/analytics/index.php and Actions > Export "Students at risk of dropping out" model
      10. Open the downloaded file and note the number of rows
      11. It SHOULD be higher than the previous number of rows

      12. php admin/tool/analytics/cli/evaluate_model.php --modelid=XXX (where XXX is 'Students at risk of dropping out' id)
        		 

      13. php admin/tool/task/cli/schedule_task.php --execute=\\tool_analytics\\task\\train_models

      14. Go to admin/tool/analytics/index.php and Actions > Export "Students at risk of dropping out" model
      15. Open the downloaded file, the number of rows SHOULD be equal to the previous one but higher than the former one (here we are testing that evaluation files are not merged into the exported file)
      Show
      Here we are going to test that: Multiple training task executions result in multiple training files generated. These file contents can be combined into 1 and downloaded You need 1 course with a past start date and end date and with at least 1 student Students at risk of dropping out models should be enabled and should have a time splitting method defined Moodle will only use courses for training if they have a minimum of activity (more than 10 logs per student in a course) You can generate all these logs or just edit lib/classes/analytics/target/course_dropout.php is_valid_analysable and make it return true php admin/tool/task/cli/schedule_task.php --execute=\\tool_analytics\\task\\train_models Go to admin/tool/analytics/index.php and Actions > Export "Students at risk of dropping out" model Open the downloaded file and note the number of rows Enrol another 2 students to that course php admin/tool/task/cli/schedule_task.php --execute=\\tool_analytics\\task\\train_models Go to admin/tool/analytics/index.php and Actions > Export "Students at risk of dropping out" model Open the downloaded file and note the number of rows It SHOULD be higher than the previous number of rows php admin/tool/analytics/cli/evaluate_model.php --modelid=XXX (where XXX is 'Students at risk of dropping out' id) php admin/tool/task/cli/schedule_task.php --execute=\\tool_analytics\\task\\train_models Go to admin/tool/analytics/index.php and Actions > Export "Students at risk of dropping out" model Open the downloaded file, the number of rows SHOULD be equal to the previous one but higher than the former one (here we are testing that evaluation files are not merged into the exported file)

    Description

      As discussed with Damyon, Tom and Elizabeth, we are not yet ready to work on ensemble models and we will not have time to release moodle with trained built-in models.

      Data gathering is not going awesomely well either so it would be nice to find a mid-point to allow people to easily share training data with us. The proposal is to add a new action to non-static models to export all their training data into a file.

      This file would merge all training files that were generated during the different cron runs into one and stream it to the admin / manager web browser for download. We will later need to establish a manual protocol to allow people to upload these files to Moodle HQ. These files should be very big as they will only contains numeric values.

      Attachments

        Issue Links

          blocks

          Improvement - An enhancement to an existing Moodle feature. MDL-59067 Calculate indicators way before the target is available

          • Major - Major loss of function, incorrect output
          • Closed
          caused a regression

          Bug - A problem which impairs or prevents Moodle from functioning correctly. MDL-60375 Analysable-level files always stored into unlabelled filearea

          • Blocker - Blocks development and/or testing, prevents Moodle from running (applicable to bugs only)
          • Closed

          Activity

            People

              dmonllao David Monllaó
              dmonllao David Monllaó
              Simey Lameze Simey Lameze
              Eloy Lafuente (stronk7) Eloy Lafuente (stronk7)
              Jake Dallimore Jake Dallimore
              David Woloszyn, Huong Nguyen, Jake Dallimore, Meirza, Michael Hawkins, Raquel Ortega, Safat Shahin, Stevani Andolo
              Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved:
                13/Nov/17