Canvas X: Aug 2019 - Update

[dnehring]

Hello all! My name’s Daniel Nehring and I’m on the data science research team here at Instructure.  We’ve been busy behind the scenes working on improving our Nudge project that was implemented last year as a part of Canvas X.  

We are not currently seeking any more volunteers for Canvas X for the Fall Aug-Dec 2019 semester, however, I just wanted to create a quick update for the community/interested parties. 

Nudge is a prototype service that helps students effectively manage their time and coursework. Nudge currently messages students via canvas conversation messages. In this pilot students have the ability to opt-out of receiving nudges and provide feedback whether the nudge was or wasn't helpful. 

When enabled, Nudge sends the following messages to students through Canvas:

• Upcoming Assignments: An assignment is due soon and the student hasn’t turned it in yet. Prompt them to turn it in / finish.
• Late Assignments: An assignment due date has passed and the student has not turned it in. If the assignment has an applicable “until date”, we prompt them to submit late.
• Positive/Generalized Nudges*: This category is more  broad, at the moment we will be sending general study habits, encouragement, and praise.   

Nudge leverages machine learning to determine if/when to send a Nudge.  This is a departure from a threshold/rules based approach. So instead of setting a rule “send message x hours before an assignment is due," our new model takes in a series of inputs and outputs either an optimal date/time or a “no nudge”.  This is significant because we do not to send too many Nudges to students, which could lead to students ignoring Nudge altogether. Conversely, we do not want the students to become reliant on the Nudging system and use it as another assignment reminder system. 

We're pretty excited about Nudge and If we have any further updates we'll posting them here in the Community page.

Privacy Notice: Canvas is committed to keeping you and your student’s personal information private. All participants will be able to opt out at any time. Any and all use of the data from this experiment will be used to make Canvas a better product and not shared publicly without express prior consent.