Instructure Research: Look Ahead

It’s an exciting time to be working at Instructure. With a new focus on analytics and the predictive tools that are being built and tested over at Canvas X, there is a new drive to leverage data and the sky is the limit. We are dreaming up simple yet powerful tools that assist teachers and students with the logistics of participating in a course. This helps our users focus on what’s most important: learning.

In the same spirit of the tools offered in Nudge and the On-Track Predictor we want to provide intelligent suggestions to help students stay aware and teachers to intervene when needed. With that in mind I wanted to take a stab at a new project. For the past few months I have been trying to prove the viability of a model that can predict late submissions for current assignments. I am happy to report that this model appears to be viable and is something that we may implement down the road. 

At the end of this preliminary finding there are two stand-out components in the model that I feel are worth mentioning:

1. A student’s punctuality score (the average time that work is submitted before/after the due date)

2. Percentage of class submissions that are “late” during the current term.

These two variables offer some insight into the potential uses of this model. From a student perspective the model could provide an alert when it detects an upcoming “late” submission. Not only will this be a helpful reminder but it can also boost a student's accountability and ownership of their work and course performance.

From a teacher's perspective, the model may help prompt student intervention before a problem arises. A gentle nudge in the right direction is always a better conversation than a course correction after a missed deadline. The model may also alert a teacher when it detects many students at risk of missing an upcoming due date. With this type of foresight we hope to assist teachers in dynamically adjusting the pace of their course before running into any pitfalls that may stall course progress.

While we are still making tweaks and updates on this model, the two components mentioned above are simple concepts that may be applied to our Analytics Beta and Nudge tools in the near future. Keep an eye out, keep sending us your input, and keep learning!

 

Stephen Marshall
Data Science Intern, K-12 Customer Success Manager