Uploaded image for project: 'Moodle'
  1. Moodle
  2. MDL-59063

Add a report showing which analysable elements are not valid for each of the system prediction model

XMLWordPrintable

    • Icon: New Feature New Feature
    • Resolution: Fixed
    • Icon: Major Major
    • 3.4
    • 3.4
    • Analytics, Reports
    • MOODLE_34_STABLE
    • MOODLE_34_STABLE
    • MDL-59063_master
    • Hide
      1. You will need a site with a few courses and a few users:
        1. C1: A course with a future start date
        2. C2: Another course with past start date and no students
        3. C3: Another course with past start date, students and no course end time
        4. C4: Another course with past start date, students and a past end date (less than 1 year between start and end date) and no students activity between those dates
        5. C5: Another course with past start date, students, a future end date (less than 1 year between start and end date) and students activity (quickest option is to enrol just 1 student, log in as that student, access the course main page and press refresh like 20 times)
      2. Create a few more courses with default values to test pagination; you can create them with php admin/tool/generator/cli/maketestsite.php --size=S --filesizelimit=1 it will be fine because end date is not set when creating courses this way and we only want to test that pagination works, we are already testing other stuff usign the courses above
      3. Go to admin/tool/analytics and select "Invalid site elements" for "Students at risk of dropping out" model
      4. This list here below shows what should happen for each of the courses described above
        1. C1 should be listed, "The course is not yet started" message in both columns
        2. C2 should be listed, "No students" message in both columns
        3. C3 should be listed, "The course does not have an end time" message in both columns
        4. C4 should be listed, "Not enough course activity between the start and the end of the course" in "Invalid for training" and "Course already finished" column in "Invalid to get predictions" column
        5. C5 should not be listed
      5. Check that pagination (only previous and next buttons) works as expected and there are no duplicate results in different pages
      Show
      You will need a site with a few courses and a few users: C1: A course with a future start date C2: Another course with past start date and no students C3: Another course with past start date, students and no course end time C4: Another course with past start date, students and a past end date (less than 1 year between start and end date) and no students activity between those dates C5: Another course with past start date, students, a future end date (less than 1 year between start and end date) and students activity (quickest option is to enrol just 1 student, log in as that student, access the course main page and press refresh like 20 times) Create a few more courses with default values to test pagination; you can create them with php admin/tool/generator/cli/maketestsite.php --size=S --filesizelimit=1 it will be fine because end date is not set when creating courses this way and we only want to test that pagination works, we are already testing other stuff usign the courses above Go to admin/tool/analytics and select "Invalid site elements" for "Students at risk of dropping out" model This list here below shows what should happen for each of the courses described above C1 should be listed, "The course is not yet started" message in both columns C2 should be listed, "No students" message in both columns C3 should be listed, "The course does not have an end time" message in both columns C4 should be listed, "Not enough course activity between the start and the end of the course" in "Invalid for training" and "Course already finished" column in "Invalid to get predictions" column C5 should not be listed Check that pagination (only previous and next buttons) works as expected and there are no duplicate results in different pages

      This new report for admins / managers / researchers would show, for each prediction model in the system, which analysable elements (what is an analysable element depends on the model, it can be a course, a user...) can be used by the prediction model and which of them can not.

      May be easier to explain this using students at risk prediction case. If enrolments and / or course start and end dates are not properly set the course can not be used for training. The target is the one that will discard elements for training and for prediction. This is about calling each of the site analysables (courses in students at risk prediction as the analysable in use is student enrolments) is_valid_analysable method both for training and for prediction.

      During today's analytics meeting we expanded this a bit. The idea is to add an option (new method returning true/false) to targets that, if enabled, would generate an insight for the users with moodle/analytics:listinsights capability on the analysable context about what should they do if they want to get xxx model predictions.

      This could use target's is_valid_analysable and is_valid_sample methods

            dmonllao David Monllaó
            dmonllao David Monllaó
            Jun Pataleta Jun Pataleta
            Andrew Lyons Andrew Lyons
            Andrew Lyons Andrew Lyons
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated:
              Resolved:

                Error rendering 'clockify-timesheets-time-tracking-reports:timer-sidebar'. Please contact your Jira administrators.