Shared modeling can help schools predict, avert dropouts

A research team co-led by Cornell found that for schools without the resources to conduct learning analytics to help students succeed, modeling based on data from other institutions can work as well as local modeling, without sacrificing fairness.