Predictive data platforms spot at-risk students earlier, helping administrators, teachers, and parents ensure every student graduates ready for their next step.
Predictive analytics is the practice of using software to identify and visualize complex relationships in the real world. In education, we often look for key events that have a statistical relationship with a major outcome in historical data. We then use that statistical relationship to predict future outcomes for current students. For example, we might use 9th-grade algebra outcomes to predict on-time graduation. Often, we’re looking for multiple variables, like Johns Hopkins’ ABC framework, and part of the work is to understand how variables relate to one another. For example, using student demographic and course history data to estimate equity disparities in future AP/IB enrollment. This sort of analysis is useful for 1) measuring existing disparities as well as 2) focusing staff attention reducing those disparities by using actionable data.
Why should we care? I believe there are three main arguments to use predictive analytics. First, it is impossible to process all of the data in front of us. Machines can help us with that task by distilling large data sets down to just a few data points. Secondly, a computational approach will likely be more precise than other methods. Better precision means fewer errors when assessing needs and assigning resources. The third reason supporting predictive analytics is that they can focus of attention on the things the organization values most. This alignment between priorities and technology is invaluable for school systems.
Understandably, many people are excited about predictive analytics and they should be smart about how to use them. For example, a recent New York Times Magazine article highlights how Pittsburg officials have augmented traditional screening systems with machine learning predictive algorithms. In this case, the article highlights how the algorithm leverages data about the family rather than just about the specific child in question. But when statistical relationships are noisy, predictive estimates are susceptible to errors too. Pittsburgh is getting it right because they are careful in how they use predictive data to augment rather than supercede human judgment. Education needs to do the same thing. See the white paper by the Metrolab Network for a larger consideration of ethical guidelines.
Predictive analytics represent a new toolset, but the real work of solving problems and adapting to our challenges is accomplished by people. When we implement predictive analytics, we need to make sure that we are empowering people to use their experience and skills to the best of their ability.
Designed to improve long-term student retention and graduation rates, Hoonuit’s Early Warning solution uses predictive analytics based on established and proven research to streamline the execution of intervention strategies through a user-friendly interface. Our solution also guides educators in identifying at-risk students earlier and integrating the intervention workflow to easily and consistently track the progress of each implementation. Learn more here.