Integration and Predictive Analytics: Why We Need to Analyze, Predict, and Act
We are all machines. Our brains are the ultimate technological marvel, effortlessly performing tasks that confound current technology. The human brain is a finely tuned and integrated solution, collecting data about the world from all of our senses and producing both reaction to the stimuli and prediction about what comes next. It is the ultimate “Integration Engine.”
The struggle to effectively integrate data was reinforced in my previous job. In a school district, during a presentation to our board of trustees, one trustee requested that we implement a student achievement “credit score”. Basing the idea off of the consumer credit score, it would reflect data about a student, measured against the performance of others with similar academic and behavior traits. This, in turn, would assist in the determination of intervention strategies. It was an interesting and rational approach based on the volume of information we had about students and the potential to predict what would happen and intervene where necessary.
Our barrier to implementation was not conceptual or ideological. It was technical. That is where many school districts struggle today. Gathering the data necessary to make this happen was a herculean task, requiring countless hours of manual data management. This reality, common in schools today, reflects an understanding of the value of data and its predictive capability. Collecting meaningful data and forming it into something actionable is the practical step that has eluded us. What I believe is missing in this process is an understanding of how to bring solutions together, identify the right data at the right time, and frame the results in terms of student growth.
How do we do this? First, we must start by shifting our data conversation to one of quality over quantity. Too often we are focused on data-driven decision making, and that focus leads us to gather all the data we can, from all the possible locations it resides, and to put all of that in a data warehouse and dashboard solution that, from the outset, confuses and complicates the issue. I have personally been in environments where district leaders are presented with lengthy and data rich spreadsheets, with very little to connect the numbers on the page to real-world results. It is one thing to know the standardized test scores of every student in the school for the past ten years. It is another to filter that down to just the information that allows us to make a decision. One is reflection, the other is action.
Be careful in your analysis, ensuring that only quality data that matches your purpose is used
We need to move the conversation about data collection away from data-driven decision making to decision-driven data collection. Let us first figure out why we collect data, the outcomes we are striving for, and then go out and assess and gather that information. Let us populate systems with meaningful data that allows educators to focus on results, rather than on becoming data scientists, which is what most solutions encourage by default.
Once we collect the right data, we first have to understand and address the challenge of data distribution. While all schools have centralized ‘master’ data systems, think SIS and ERP solutions, these systems do not house all data used in schools. Some systems attempt to be ‘all in one’ solutions, but more often schools look for best-of-breed applications, or solutions that have been vetted to meet their unique needs. Often, disparate systems are used, and each system carries a separate set of requirements for data capture, storage, and management.
Solving this problem has been an early unachievable dream for decades. Many well-meaning initiatives have been advanced to try and solve this, from state-level data management systems to a variety of protocol (or data language) definitions. Nothing has transformed the landscape yet. Other fields have solved this problem already, and have both standards and technologies in place to ensure that data moves elegantly, reliably, and securely between systems. Do a simple web search for “integration engine,” and you will see all the other industries that have figured out how to standardize the exchange of data between systems. Lacking a common approach in education is more than unfortunate; it is borderline irresponsible, given the potential for data to inform quality work with students. After we collect the right data and foster it into an ecosystem of free data exchange, we can begin to use it in meaningful ways. Analytics is easy, or at least relatively easy, compared to the job of transforming analytics into something with real world meaning. Showing data that reflects where we have been is standard fare. I would argue that this is one of the earliest drivers of technology in schools, and that it remains the hallmark of much of what is sold. We are surrounded by data, and almost all of it is reflective.
According to former Delaware Secretary of Education Mark Murphy, "At the end of the day you can build the systems, and that is all really important work, but if you do not have people, who have great capabilities in how to use that data and how to turn that data into usable formats for educators and policymakers, then it will just live and die in that database and not actually inform policy, not actually inform practice." This is where we can place value in education “data”. It is crucial to look back, evaluate our past practice, and learn from it. For today’s students, we are afforded no luxury to take our time and assess long-term historical trends. Instead, we need to cast our glance forward, to mine data that is connected to our real questions and goals, and to begin making predictions. Will a change in this reading program increase comprehension levels amongst our most remedial students? Will that after-school program reduce in-class behavior issues at our middle school? This is what we want to know, and what is responsible to ask, for today’s students.
In the end, we must follow the mantra: analyze, predict, and act. Be careful in your analysis, ensuring that only quality data that matches your purpose is used. Use decision-driven data collection. Then integrate that data with information from a variety of sources. Look for easy solutions here-openness, flexibility, and an intrinsic ability to work in your systems.
Predict the outcomes; attempt to understand how your decisions today affects the students you have now. Be intelligent consumers of predictive analytics, using them as one tool in your toolbox for creating a better learning environment. Use the power of the vast amount of data you have to better understand the choices you make.
Finally, act. Do something for your students now, and do not wait for historical data to tell you what to do with the next generation of students that pass through. Lead with passion, intention, and purpose. Leverage the great work you have done to find the right data, bring it together, and draw predictive conclusions about your students’ work. It matters to them as much, or more, than it matters to you. In the end, create an ecosystem that encourages this kind of quality throughout the organization, focused on conversations informed by the right kind of data. By doing so, you will find that a difference is made today, for the students you see in the halls right now.