Our final session highlight is ‘Lessons from Deploying In-house Predictive Retention Analytics’ by Jeremy Anderson, Rich Silva, and Ashley Muraczewski at Bay Path University.
Normal reporting is after the fact and while it tells a story, it can take a lot of time for the data to be available and by then, it is frequently too late to help those who make up the story. Predictive analytics are attractive as they can be used much earlier in the story (i.e. via interventions) and strategically applied, can change the story before it ends. However, predictive analytics can be difficult and expensive to implement and thus institutions can struggle to both initiate and leverage those programs.
Bay Path prides itself on serving underserved populations. However, they didn’t view access alone as a definition of success. It was also about persistence and graduation and they saw predictive analytics as a way to make a difference in those areas. Like most institutions, there was an abundance of anecdotal evidence concerning student success. Utilizing data that was already in their warehouse and existing analytic tools, they embarked on a collaborative, holistic effort with the campus areas involved in student success. They chose a contextual approach that included measures that were both based on the cohort as well as the individual student. While they found many variables that affected student success, their final product focused on 12 that proved to be truly predictive. While the initial effort focused on undergraduates and the interventions are manual, they have plans to spread the effort to the graduate and doctoral populations and are looking to automate the interventions via their CRM system.
To learn more about how you can initiate a successful predictive analytics program at your institution, please register to attend the 2021 HEDW Virtual Conference.
Our next session highlight is ‘DIY Data Applications – Self Service at its most basic level’ by Kristin Kennedy at Arizona State University (ASU).
Business Intelligence (BI) continues to evolve at a rapid pace. User’s expectations continue to change and grow and the amount of available data is growing as well. Given that resources continue to be tight, the situation is causing most BI operations to rethink how they service their users while continuing their efforts to become a true data informed institution.
As ASU examined their situation, they found that many users already had powerful data tools that they used every day and in some cases had their own datasets. They decided to embark on a partnership approach that focused on self-service. By partnering with customers and power users they were able to help them better leverage the tools they already had, increase utilization of existing data, become more self-reliant in meeting their reporting needs, and also adopt better data management strategies for their datasets. The program now includes an on-campus data conference and training program that includes badges to indicate proficiency. The session will include some valuable lessons learned (such as how to implement a self-service program with an incredibly diverse set of user skills) as well as some of the training materials that they use in the program.
To learn more about how ASU designed and implemented a successful DIY program, please register to attend the 2021 HEDW Virtual Conference.
The third session highlight is ‘When Regression Isn’t Enough: Using Simulation to Make Predictions’ by Abby Kaplan at Salt Lake Community College (SLCC).
Predictive analytics is an area of BI that continues to garner considerable interest in Higher Education. The events of the past year have only heightened that interest as institutions try to understand what changes they need to make and how those changes can impact the future. One of the challenges of predictive work is understanding the available data. There are different methods of performing predictive analytics and having a clear understanding of the situation is crucial in choosing the right path.
SLCC’s situation involved a placement test. A unit was considering changing the design of their placement test to improve how well it predicted what courses students should take and how well they might perform in those courses. In looking at past data, the available data was not suitable for regression modeling. The team turned to simulation and by using R programming, predicted how changing the design of the placement test might affect the outcomes. The case study presented here will include some valuable lessons in doing this type of work. Those lessons include what situations are good candidate for simulations; the importance of keeping the situation manageable; understanding that the results are not an absolute, but can provide valuable guidance; and being sure you have a clear sense of the outcomes.
To learn more about SLCC’s experience with simulation, please register to attend the 2021 HEDW Virtual Conference.
The second session highlight is ‘From Data Governance to Data Strategy: The Evolution of the Institutional Data Governance Program at the University of Toronto’ by Kiren Handa and Jeff Waldman.
Data governance programs are always challenging to build and fully implement, and can be incredibly complex as they typically touch all areas of an institution. Success is frequently achieved incrementally and requires navigating both cultural and structural issues. Senior leadership at the University of Toronto made the commitment to data-informed decision making and recognized that data governance was a cornerstone of that process. Having a highly decentralized structure, they formed a committee of interested parties across the institution and began the journey. Working together and consulting with the various constituencies across the institution, they focused on existing issues and what was desired or expected from data governance. They were also able to leverage on campus resources including faculty and campus partner expertise.
The committee knew they had to get off on the right foot which meant being deliberate, inclusive and methodical to ensure institutional-wide support for the program at its onset. They used the results of the consultations to evaluate the current state, build the case for change and articulate a vision. The culmination of the process led to publication of their foundational documents as well as what is being labeled as a set of ‘flagship’ initiatives. These initiatives build the foundation necessary for the institution and formulate its data strategy. The strategy aims to achieve the governance vision and involve transformations relating to people, process, technology and policies. To learn more about their journey, please register to attend the 2021 HEDW Virtual Conference.
New for the 2021 HEDW Virtual Conference is session highlights. Each week prior to the conference, we will be highlighting one session and providing additional insight into what that session will cover. To start, we are focusing on ‘The Best of Both: Cornell’s Hybrid Data Warehouse/Data Lake Architecture’ by Jeff Christen at Cornell University.
One of the challenges in the BI world these days is navigating the sea of technologies, terminology, and concepts that we are regularly bombarded with. Even more daunting is that many of these things do not necessarily have a clear, standard definition that transcends every conversation. With this backdrop, Cornell embarked on the journey to determine what data architecture would meet their current data reporting and analytic needs and enable them to meet future demands. This session chronicles that journey and where Cornell eventually ended up.
To start, they asked the question, did they need to build a new analytic platform? They had an established data warehouse that had served them well in the past, but as the data landscape was changing, could it keep up? And if they decided to build a new analytic platform, should it be another data warehouse or perhaps a data lake? While these types of conversations frequently focus on strengths versus strengths, Cornell found that the conversation was just as much about technology versus methodology.
In the end, they chose a hybrid solution that they are finding provides the flexibility their users are asking for while also allowing for data discovery. To hear the details about the journey, please register to attend the 2021 HEDW Virtual Conference.