Aggregate Student Data for Better Decision-Making
With the shift to remote learning, online views into student performance are more important than ever. In a classroom, teachers can assess who needs help and who needs to be challenged every day. Without the in-person interaction that a classroom affords, these kinds of assessments are not so easy. Many school districts are turning to data and analytics to help inform their decisions to ensure each student is challenged and engaged.
A client focused on the K-12 marketplace used three different online applications to build student reading and vocabulary skills. In addition, teachers uploaded data from offline assessments on Lexile levels or GRL (Guided Reading Level) at key points in the academic year. The client needed a way for teachers to be able to view aggregated information about this data in a single location for a more complete view of student progress on an individual and combined class level basis.
They wanted a solution that would provide the teachers with the ability to review individual student data in aggregate across all the learning tools to inform decisions and monitor student progress. Their goal was to allow teachers to auto-generate student groupings or create custom groups to align students with similar skill sets based on a variety of different factors such as current reading level, interests, or skills that require additional development to meet grade-level expectations. Having tangible data and analytics would take the guesswork out of student alignment and organization to allow teachers to spend less time on classroom organization and more time on instruction and individualized attention.
They needed a technology partner who could work collaboratively with them to design and implement a solution.
Unicon’s experience with data and analytics made us the right partner for the job. Extensive knowledge in creating centralized data repositories as well as our experienced design team that is skilled in collaborative development made Unicon well positioned for this work. Our teams worked closely with the business stakeholders to understand their overall vision and develop the right solution.
Reaching the overall product vision for this solution started with a unique brainstorming and prototyping session with key business stakeholders, technical engineering, and K-12 teachers providing feedback on the designs. The project kick-off team was tasked with building rapid prototyping examples of a comprehensive analytics dashboard during a 2-day internal workshop. The top designs and concepts were then taken into the field to be tested with actual teachers whose reactions and recommendations were incorporated back into the designs and dashboard requirements. This type of forward-thinking and planning allowed the project team to get feedback during project initiation to validate assumptions and make adjustments before actual development started.
For project execution, Unicon partnered with the client to form a blended Agile team. The team used SCRUM and followed standard practices for Sprint Planning, Sprint Reviews, and Retrospectives. Engaging stakeholders in reviews and demos and maintaining a dynamic product backlog allowed the team to react to changing priorities based on stakeholder and beta client feedback. Two phases were outlined for this project; a beta release and a production release. Collecting feedback back from beta customers throughout the project proved to be crucial to overall project success and future adoptions. After the beta release - the official release of this project is fall 2020.
The technologies used to enable the teacher dashboard focused on event streaming and development of a data lake. Each application the students used was modified to send IMS Caliper events to a microservice REST API. These events were captured and delivered to an Amazon Web Services (AWS) Kinesis stream. Kinesis enabled the storing of the events in an AWS S3 bucket, for use with other data lake technologies in the future. In addition to storing the events, the event data was also parsed and stored in a more traditional relational database (RDBMS). AWS Lambda functions were used to get the data from Kinesis to the relational database. By streaming events real-time from each application to a central data lake, the teacher dashboard did not need to communicate individually with each application. The student data was provided to the browser page through normal microservice REST APIs. Each of these APIs leveraged the relational database where the event data had been stored.
To visualize the trends in the data lake, a React and eCharts single page application was created to surface class level trends to instructors. Each trend is focused on a class’s progress in a single week. Instructors can drill down in the data, seeing how individual students and groups are progressing, and also view historical trends. The application is hosted using AWS S3 and Cloudfront and pulls information from the unified APIs provided by the data lake. React and eCharts provide a platform for quickly designing new visualizations, user testing the visualizations with a group of instructors, and gives fine-grained ways to refine the presentation of trends for general consumption.
Benefits to Instructors and Learners
While the value of the effort to teachers was clear to the team when the project started in the spring of 2019, the rapid shift to online learning for K-12 classes in the spring of 2020 due to the COVID pandemic only reinforced and amplified the overall value proposition. The usage of the three student applications feeding the data lake and populating the teacher dashboard spiked as learners began accessing the systems from home on a daily basis to continue the development of the literacy skills that they could no longer address in class. Teachers already overwhelmed with technology demands for the move to full-time remote learning will benefit from having a single location to view student performance data and make decisions that will improve their learning experience.