A common area for learning analytics is to look for ways to support struggling students. Largely, the focus has been on how to identify students of concern. In a free and open report, a collaboration of researchers from major Universities have outlined exactly that: how to identify students who may not graduate high school.
Here are the amazing findings of a dataset of 11,000 students:
- Machine Learning far outperformed the human method for identifying at risk students
- The Predictive ML Model only got more precise as time when on (and more data collected) as opposed to the human method that became less precise.
- The ML models aims to rank students by order of predicted need for intervention and to also determine when the intervention is needed
- The ML model also aims to predict the likely success of a given intervention.
Powerful stuff! It’s a quick and worthy read for any Education Data aficionado.
Disclaimer: The data and graphics used on this site are simulated re-creations intended to protect the privacy of the original data sources.