A hands-on approach to predictive analytics provides both the data and support for solutions. Whether higher-ed is using predictive analytics to identify students who are at-risk of dropping out or finding new data-based enrollment strategies; software output alone is not enough. No matter the application, synergy of higher-ed specialists and analytics providers is key to implementation of data-driven solutions.
Predictive analytics is a relatively recent approach to find admissions and student retention solutions. Most universities do not have in-house data scientists to aid in implementation of predictive analytics solutions. Consequently, universities are left to ponder the output from black box software solutions.
Each university is different. Therefore, each solution should be tailored specifically to that university. Additionally, VP’s in charge of applying findings, have practical questions about their output. Customer service is not equipped to answer these questions. Experts involved in the analytical process have answers to in-depth output questions.
The Human Role in Predictive Analytics
The human role in predictive analytics and machine learning is critical.
So, analytical technology provides a basis for data science and higher education practitioners to partner, and create new strategic plans for students and the university. The human element is critical to the entire operation.
Furthermore, creating student intervention and communication plans exemplify the importance of human interaction with predictive analytics. One of the topics we speak about frequently is segmenting students who are at-risk of dropping out into similar groups for intervention planning. By placing at-risk students into groups based on various qualities, we can develop a consistent intervention plan for each group rather than a completely individualized intervention for each students. This helps the university use student success resources efficiently and intelligently. Human interaction with the analytics process is critical when determining:
- How many intervention groups should we develop?
- What should the maximum size of each student group be?
- What qualities do these students have in common where the segmentation algorithms decided to group these students?
- What intervention resources would help this group of students?
- Do we have enough of a specific resource to serve the entire group?
- How should we approach communications and messaging with these groups in a positive way, based on their qualities?
Another issue arises as the university makes changes based on predictive models. A shift in the business rules, such as allocating new scholarship amounts or new scholarships all together, can change the entire enrollment dynamic. This is where adaptive management or adaptive analytics is critical. The impact of new initiatives must be quantified, and models must be updated. Most importantly, human interaction is imperative for proper interpretation and to ensure that new applications have not exceeded the capabilities of a predictive model.
Collaboration Between Data Science Experts and Higher-Ed Specialists
Simply handing off a list of students who are predicted to drop out through software is not helpful. The university requires and deserves a larger picture. Thus, the data scientist should provide baseline suggestions for intervention based on the university’s available resources. Furthermore, higher-ed specialists bring a unique perspective to these challenges. They have the most hands-on experience with their students and the best insight into the students’ struggles. Bringing this experience in to the conversation guides the analytics problem-solving process. These experts formulate the questions that need to be answered with analytical insight.
In Conclusion, universities need consistent interaction with analytics providers for the most impact from analytical investments. This means a finding data science solution that combines advanced analytics software, with consulting time where both parties can collaborate, answer questions, and create new solutions to better support their students. Implementation of a hands-on approach to predictive analytics involving data scientists and higher-ed specialists will provide the university with optimal data-based practical solutions.