“Predictive analytics” is the latest buzzword — but what do predictive analytics mean for teachers, administrators, and schools?
Imagine a world in which educators had the ability to pinpoint exactly which kids are at risk of failing their exams at the end of semester, which interventions have the greatest impact on student achievement, or which students are unlikely to graduate several years from now.
It may sound like science fiction, but in reality, many industries have been using predictive insight with relative accuracy for years. Stock market analysts can predict when the market will rise or fall with a fairly high level of certainty, pharmaceutical companies can predict whether a patient is likely to adhere to a prescription, and retailers now know how much an individual customer is likely to spend throughout their entire lifetime.
Thanks to emerging technology, today’s educators are similarly positioned to leverage the power of predictive analytics — to the benefit of schools, districts, and the community at large.
These visualizations of data modeling look complex, but understanding predictive analytics doesn’t have to be. Below, I’ll break down predictive analytics and how they’re applicable to educators.
Predictive analytics combine statistical approaches such as correlation, regression, clustering, and advanced use of other statistical techniques (i.e., machine learning) to examine relationships among the data and estimate the likelihood of future outcomes. Using predictive modeling, an organization can measure and visualize predicted outcomes and align decision-making to positively influence those outcomes. In simple terms, predictive analytics techniques reveal patterns in data to highlight relationships that are actionable.
An example of this is the emerging adoption of predictive early warning systems. Such systems model the relationships between grade point average, credit earnings, behavior incidents, assessment results, and attendance and high school graduation outcomes (students with observed outcomes). These modeled relationships allow school districts to accurately predict the likelihood of graduation for current students (students with unobserved outcomes). With a finite amount of time and resources, there is a distinct advantage in being able to provide supports sooner rather than later.
With clear potential for district-, school-, and student-level improvement, it’s easy to understand why predictive analytics are quickly becoming a fixture of the educational technology landscape. In fact, many districts are already making use of predictive capabilities to improve decision-making processes.
Predictive analytics typically use factors such as course grades, attendance, suspensions, test scores, program assignment, school enrollments, and extracurricular involvement, but it’s important to point out that some tools also allow educators to use qualitative methods as classroom observations, surveys, and interviews as well. Depending on what you are looking to predict — at-risk students, college and career readiness, or future school enrollments — combining predictive analytics with qualitative data will almost certainly enhance strategic decisions. In other words, the most effective use of predictive analytics marries the science of data analytics and the art of decision-making.
Predictive analytics deliver important insights earlier than ever before, giving you the necessary data to understand your achievements and challenges within the context of a continuous improvement cycle. Having the data at your fingertips allows you to focus on finding solutions, capitalizing on successes, and empowering your decision-making to funnel resources where they’re needed most.
Predictive analytics have countless applications in educational settings — from predicting individual students’ future academic success to individual student attendance. Predictive tools are frequently used for classification (identifying at-risk/not at risk), but they can go well beyond binary outcomes. They also provide great detail about each student’s level of risk and the factors contributing to their struggles.
The use of predictive analytics is not limited to struggling students. Predictive tools can also identify which students are likely to enroll and/or persist beyond their first year in college. And this is just the beginning for predictive analytics. If the past few years are any indication, their applications are as limitless as the questions we’re able to ask of them.
For example, intuitive dashboards designed using output from predictive analyses can tell educators whether a student is likely to graduate high school but is not prepared to succeed in college or the workforce. Similar techniques can be used to forecast student growth on a variety of measures. Aggregating these measures by classroom, grade, school, and/or district can provide local education agencies (LEAs) on what their proficiency status is likely to be in the future.
In combination with data from multiple sources (i.e., birth rates, city planning data, and historic school enrollments), predictive modeling techniques can forecast enrollment trends with a high degree of accuracy. School LEAs are able to accurately forecast school enrollment trends over the next five years or more. Other methods are used to identify ideal hiring pipelines and measure the impacts of programs and interventions. In another type of analysis, predictive analytics can be used to cluster similar schools together on a variety of school characteristics. School LEAs are using the resulting clusters to help educators design professional development efforts and facilitate inter-school sharing of effective practices among school staffs within similar contexts.
Of course, schools can’t take advantage of the power of predictive analytics without a consolidated and comprehensive data infrastructure. That’s why, as predictive technologies continue to advance, it’s essential that districts cultivate a data-first culture among educators and administrators. While predictive analytics are by no means an infallible crystal ball, they empower educators to deliver additional support to students earlier — and in a more precise way — than previously possible.
At Hoonuit we embed predictive analytics throughout our solutions to enable educators to make the best decision for their schools and students, and refine those analytics as new technologies emerge. Learn more about our data analytics solution here.
UPDATED 10.30.18: This post was edited to remove ambiguity between applied predictive modeling and modeling for causal inference.