Managing the Student Success System

Administrators are responsible for provisioning the Student Success System by: managing target courses, building predictive models, and monitoring active courses.

  1. Adding and deleting target courses
  2. Configuring a predictive model
  3. Updating a predictive model
  4. Using predictive models
  5. Running a simulation

Adding and deleting target courses

Add a target course

  1. On the Student Success System Administration page, click Add Course.
  2. Select the course you want to add from the Select Courses window.

Delete a target course

On the Student Success System Administration page, click Delete from the context menu of the target course you want to delete.

Configuring a predictive model

Once you add a target course, you can configure a predictive model for it. While the model is being built, the progress status displays as "Building [ ]% complete." Click Refresh to refresh the building status or Cancel to cancel it. Once the build completes, the model status displays as "Ready" and the Mean Squared Error (MSE) displays. The MSE is also visible on the Review Model page for the predictive model.

MSE is calculated as the average deviations between the estimated grades used in calculating the success index and actual grades. The average is calculated over all weeks and all data domains. The value is then normalized to a percentage value. The smaller the MSE, the more accurate the predictive model is thought to be.

Configure a predictive model

  1. Click Configure Model from the context menu of the target course you want to configure.
  2. Select the domains you want to include in the model. We recommend that you exclude domains that have no corresponding data in the system to eliminate irrelevant measurements from the model. For more information about domains, see Student Success System domains.

    Note  Before selecting the Preparedness domain and building the predictive model, ensure you have Student Information System (SIS) data uploaded to the system. Contact your Student Success System administrator for more information.

  3. In the Advanced Model Options section, select the Model Aggregation Type and Data Extraction preferences.

    Note  The system selects the Domain Aggregation and Cumulative Weeks options by default. Desire2Learn recommends leaving the default selections in place.

  4. Confirm or modify the grade ranges for the three risk categories.
  5. If start and end dates are not automatically included, enter a Start Date and End Date for the course.
  6. Click the Edit icon to define which roles to include in the predictive model.
  7. In the Map Historic Courses section, click Add Course to add at least one historic course to the predictive model.

    Note  The more consistent the historic courses are with the target course you are configuring, the more accurate you can expect the predictions to be. Any historic course you add must contain enrolled participants.

  8. Click Save and Continue.
  9. On the Review page, click Build Model.

    Once you've created a predictive model for a course, it is stored in the Analytics database and generates daily predictions as part of the ETL (extract, transform, load) process.

Updating a predictive model

Once a course commences, you can no longer update its predictive model. Every time you update a model, the system builds an additional model for the course. You can use the Revision Log to switch between the various models for a course.

Update a predictive model

  1. Click Review Model from the context menu of the predictive model you want to update.
  2. Click Update Model.
  3. Make your changes.
  4. Click Save and Continue.
  5. Click Build Model.

    Note  Your new revision will appear in the Revision Log. To view the Revision Log, click Review Model from the context menu of the predictive model you want to view the log for.

Switch between versions of a predictive model

  1. Click Review Model from the context menu of the predictive model you want to switch.
  2. Click on the Set Model Version icon beside the model you want to set as active. Once the model is active, it displays a Checkmark icon beside its name in the Revision Log.

Back to top

Using predictive models

Once you've configured a predictive model, you have several options.

Review predictive model

Click Review Model from the context menu of the predictive model you want to review. This displays the model settings and revision log for the model.

Preview predictive model

Click Preview from the context menu of the predictive model you want to preview. This takes you to the class dashboard for the model.

Delete predictive model

Click Delete from the context menu of the predictive model you want to delete. This will remove the model for the course. You can re-add the course to the list at any time.

Set predictive model as inactive

Click Set as Inactive from the context menu of the predictive model you want to set as inactive. This will hide the model from instructors. The model continues to generate predictions and records them in the database. To reactivate the course, click Set as Active from the context menu of the inactive predictive model you want to reactivate.

Back to top

Running a simulation

Simulations are experiments that can be run for the purpose of testing predictive models and learning from the data they produce. You can select completed, in-progress, or upcoming courses for simulation. If you select a completed or in-progress course, you can use the simulation to observe how the predictive model would have behaved over time, view the predictions that would have generated each week for the student, and compare the simulation with how the student's actual performance in the course.

Run a simulation

  1. Click Add Course to add the course you want to run a simulation for.
  2. Click Configure Model from the context menu of the added course.
  3. Select the domains you want to include in the simulation.
  4. In the Advanced Model Options section, select the Model Aggregation Type and Data Extraction preferences.

    Note  The system selects the Domain Aggregation and Cumulative Weeks options by default. Desire2Learn recommends leaving the default selections in place.

  5. Confirm or modify the grade ranges for the three risk categories.
  6. Enter a Start Date and End Date for the simulation.

    Note  If you are running a simulation for a completed or in-progress course, your start and end date must be in the past.

  7. Click the Edit icon to define which roles to include in the predictive model.
  8. In the Map Historic Courses section, click Add Course to add at least one historic course to the predictive model.

    Note  The more consistent the historic courses are with the target course you are configuring, the more accurate you can expect the predictions to be. Any historic course you add must contain enrolled participants.

  9. Click Save and Continue.
  10. On the Review page, click Build Model.
  11. Once the model's Status displays as Ready, click Preview from the context menu of the model to investigate the simulation.

Back to top

See also

 

Desire2Learn Help | About Student Success System