top of page
  • Writer's pictureSEE Team

Predicting student dropout risks, increasing graduation rates with cloud analytics

Is it possible to predict whether students are at risk of dropping out of school? The Tacoma Public School district thinks so. Using predictive analytics tools based on Microsoft cloud technologies, the district is providing comprehensive data snapshots of student success indicators and has already helped to improve graduation rates from 55 to 82.6 percent.

Shaun Taylor: CIO

Tacoma Public Schools

And once teachers and principals can use the data model to predict dropout probability, they’ll be able to provide additional learning assistance early enough in the process to turn at-risk students around.

Not long ago, the reputation of public education in Tacoma, Washington, was as bleak as the gray skies that often blanket the city. A 2007 national study dubbed the 30,000-student Tacoma Public School district’s five high schools “dropout factories,” where many freshmen never made it to graduation. As recently as 2010, just 55 percent of the district’s high school students—well below the national average of 81 percent—earned their diplomas on time.

But administrators in the challenged district didn’t give up hope. Through their efforts, the graduation rate jumped from 55 to 82.6 percent by 2016. Today, the district is recognized nationally for its educational achievements.

So how was the dramatic turnaround achieved?

It started with a radical vision: What if teachers and principals had analytical tools to look at all the data surrounding a student and could then predict whether or not a student was likely to disengage and ultimately drop out? Armed with such tools, educators and administrators would be empowered to reverse the trend and help more students succeed.

“We have so much data in the district, and we wanted to get better at analyzing trends of student success,” says Shaun Taylor, the district’s CIO. “By using predictive analytics, we thought we would be able to intervene earlier and work closely with those at-risk students. Then we would be able to reach our ultimate goal: getting that graduation number close to 100 percent.”

Looking to change perceptions about students

The ability to even dream about such a goal started several years ago, when new members were hired into leadership positions by the district’s board. “With the change of administration, there was a desire to be more transparent with our data, not only in the district but also in the community,” says Dorothy Kippie, Director, Technical Operations, Tacoma Public Schools. “The idea was to show people the good things happening with our students, but also to address the things that still needed work.”

Additionally, the new administration wanted to change certain perceptions within the district. “Often, students are seen as fitting certain profiles that indicate a potential lack of success, but none of those profiles are supported by analytical data. We wanted to use data to change that perception,” says Christopher Baidoo-Essien, BI Analyst, Tacoma Public Schools. “And eventually, we want to predict what the key indicators are for kids disengaging.”

Starting the journey with descriptive analytics

The district started its data-driven journey by exploring various business intelligence (BI) technologies. “We asked Microsoft and some other vendors for assistance, and Microsoft actually pointed us to a third-party solution wrapped around Microsoft SharePoint, which we were already using,” says Taylor. “That solution didn’t seem flexible enough for us, but that started a whole conversation with Microsoft Consulting Services, which helped us navigate a pathway forward.”

Microsoft Consulting Services (MCS) worked with the district to develop a data warehouse solution that captures recent data from the district’s student information system, containing student grades, attendance, health records, and other data. Using Microsoft Excel Services, administrators can view Excel workbooks in Microsoft SharePoint 2013.

The workbooks map the data to the district’s key benchmarks, such as math and reading standards, graduation rates, school environment, and readiness for life after high school. Using the solution, users have 72 different ways to view the data, with the ability to drill down to see metrics at the classroom level. “This tool was very good at helping us build some analytics to see student performance from a near-time historical perspective,” says Taylor.

Helping boost graduation rates from 55 to 82.6 percent

The Tacoma Public School School Board resoundingly embraces the value of data-driven analytics. At almost every school board meeting, benchmarks are discussed, and more importantly, so are the actions being taken as a result of those benchmarks.  Those conversations are broadcast to the community and shared across the district. In a recent Board meeting, the president of the Board, Scott Heinze, stated, “We now have this world-class data system for teachers to use. They want to know what is going on in their classrooms. “  Adds Taylor, “We finally got to the point where we had some great BI tools to see how kids were doing in the classroom, including performance, and also some different factors such as attendance, discipline, and other factors  that impacted that performance. What that did was help change the conversation for educators on how to think about where students are at and how to help them progress.”

It also helped contribute, along with educational efforts in the classroom, to the huge bump in graduation rates. “More than anything, the combination of new visionary leadership with sophisticated data analytics is helping drive improvement in graduation rates,” says Taylor.

Measuring the whole child

The district’s solution is also at the heart of its new initiative, “Measuring the Whole Child.” The program is based on four goals: helping students achieve academic excellence; creating partnerships between parents, community, and staff in educating children; focusing on early assessment and intervention to ensure academic success; and creating and maintaining safe learning environments. The program’s benchmarks use data to collectively measure a whole child to determine how well the district is helping move kids forward. “Our district is getting recognition nationally for our Whole Child initiative,” says Taylor. “Data is at the heart of making that happen.”

Taking the next step: trying to determine the probability of dropping out

Satisfied with its initial success, the district was ready to take the next logical step: using data to predict which students were at risk of not completing their education. But first, the district needed to identify the technology that could make that happen. “We’ve always wanted to use predictive analytics, but getting there has always been a hurdle for us,” says Taylor. That changed when MCS introduced Taylor and his team to Microsoft Azure Machine Learning (ML), a predictive analytics solution based on the Microsoft Azure cloud platform. “When we saw Azure Machine Learning, we started to see how it could be possible for us to realize our vision,” Taylor says.

The Microsoft Data and Decision Sciences Group (DDSG) created a proof-of-concept (POC) data model centered on Azure Machine Learning. The model intakes student data that is uploaded to Azure Blob Storage from the district’s student information system and several other on-premises information systems. Azure Data Factory enables a predictive pipeline that uses the Azure Machine Learning model created by DDSG to make predictions based on student data. Once predictions are complete, predictive results are output to Microsoft Azure SQL Database, where district board members and several IT staff members can view them using a Microsoft Power BI dashboard.

Using an accurate data model that will help predict student success

In the POC, the district used data spanning five years for students from grades 6 through 12. The data included previous demographics, and academic and student performance information to predict if a student would be at risk of dropping out the following semester. “When we started this POC, we didn’t know if any predictive analytics would be attainable,” says Baidoo-Essien. “As we progressed and used more historical data, the model proved to be almost 90 percent accurate.”

These early results gave Tacoma Public Schools confidence that it had the right tool to help it reach its ultimate goal. “With Azure Machine Learning, we proved that we have the right tool to get us where we want to go in terms of predicting student success,” says Taylor. “It’s a tool our educators will be able to use to start tackling the problem of student disengagement.”

Working toward a real-time snapshot of a student’s dropout risk

As Tacoma continues to refine the data model, the district is striving to make it more agile. “One of the challenges is that the data we looked at is historical in nature, and the data sets are semester-based,” says Taylor. “So if the data about an at-risk student is a few weeks old, that student has already lost two weeks of additional intervention and support.”

To address that issue, the district will eventually give teachers and administrators weekly reports, including factors such as attendance, the time teachers spend with individual students, and relevant disciplinary action, so there can be near-real-time indicators that show a snapshot of a student’s current risk of disengaging. “Then they could get a sense of the probability of any student disengaging in the next semester, according to that student’s history up until right now,” says Taylor.

Finding other indicators in external data

In addition, the district believes it will eventually be able to change traditional perceptions about the reasons for students’ struggles, by identifying other indicators that contribute to students’ tendencies to drop out. “There is a lot of data outside the district that we want to tap into,” says Taylor. “For example, there are a lot of government organizations that work with our students and families, and we are also interested in nutrition data and even social media. There are likely some influencing indicators in those data sets that we can bring into the equation.”

Intervening early to get students moving in the right direction

Taylor and his team are working to have a more agile Azure ML model, containing more data sets, in use by teachers and principals in the coming school year. “We’d love to have administrators be able to look into the data and see that 30 percent of a group of incoming fifth-graders need reading or math intervention,” says Kippie. “Right now, they can’t even put their finger on that. By having access to that data before the school year starts, teachers could get some additional resources to help get those students where they need to be more rapidly.”

For teachers, intervening as early as possible is critical. “When a fifth-grader, for example, goes into middle school, there’s the added pressure of changing classrooms and having different teachers for different subjects,” Kippie says. “That’s why it’s so important to help them early and move them in the right direction.”

Sharing the model with other districts

The district also wants to make sure the solution can work for other school districts around the country. “I want to make sure this solution could be packaged and shared with others so they can achieve the same level of insights,” says Taylor. “School data in all schools is similar, it’s just that the districts have different systems. By using this approach, they could feed those systems into the same data model.”

Turning a vision into reality

Eventually, the district plans to use its Azure Machine Learning solution to make its original vision a reality. The end goal is a scenario in which a teacher or principal can log into a portal each morning to see a data view of each student, and then be proactively alerted by the system when a particular student is at risk of failing a course or dropping out. “We want to make it easy for a teacher or administrator to be notified if there’s something different about a particular student from one day to the next. Once they get that alert, teachers will be able to take action and intervene, all because of the data,” says Taylor. “That’s the point we think we can get to with Azure Machine Learning.”

The district is confident that its schools no longer deserve the reputation as dropout factories. “Here in Tacoma, we have two things happening right now: we have a leadership team that recognizes the importance of knowing the truth about a student’s progress, and we have the technology tools that give us that complete picture,” says Taylor. “That gives us a lot of hope for the future that our students will become even more successful.”

Data Management and Analytics

Organizations realize a competitive edge when more employees are empowered with data. Microsoft’s unique approach to data technology delivers this capability—whether through insights and analytics or with powerful reporting for line-of-business applications. In a world where business demands the speed to compete, Microsoft data solutions cut the time it takes to go from raw data to results for everyone.



  • Twitter Clean
  • Flickr Clean
bottom of page