Interesting Patents: Oracle’s Student Retention System

Student retention system
U.S. Patent No. 11,062,411
Assignee: Oracle


Colleges and Universities constantly struggle with engaging their students in a way to keep them from dropping out. There are both a plurality of risk factors inherent to each individual student as well as external measurable parameters relating to their involvement on campus that can be attributed to a student being more likely to drop out or transfer schools. Universities provide many services to their students to promote retention such as campus-funded tutoring, freshman seminar courses, and intramural sports. The issue faced by universities today is how to identify the at-risk students in a way to be able to provide them with the established resources that would promote their retention. 

This issue is not new, however, the technology has not been able to adapt to the individual needs of each and every university. This is not for a lack of trying on the part of the universities, who for decades have been collecting retention information relating to their students. It was not until Oracle created a machine learning-based student retention system that universities have been able to use all of their historical data to identify the risk factors associated with current students. A patent issued to Oracle this week has created a system capable of implementing a machine learning network trained on prior student data to analyze and identify current students at risk without the need for a specialized or individualized system for each university.

WHat the invention would do

The foundation of the system relies on prior student data to train a machine learning model to identify the risk factors associated with dropping out and or transferring at a particular school. The system can then identify a plurality of risk factors associated with each student currently enrolled at the university from the machine learning model and assign each student with a risk score. Parameters that feed into the risk score include, “grades, involvement in university programs, social relationships, history at prior educational institutions, family educational background, similarity to other students that have dropped out, and similarity to other students that have successfully graduated from the educational institution.” 

For this system to operate effectively, prior student information is used to train the deployed machine learning model to identify the risk factors associated with a student body at a given university. The risk score can then be fed into the trained machine learning model to categorize a plurality of risk levels associated with each student by comparing them to the available machine learning model. The plurality risk categories associated with a particular student can be either aggregated or combined to display only the most relevant risk categories that, as determined by the model, would lead to the student transferring or not completing their schooling. Analytics of a plurality of students can be used to determine the highest risk factors facing the university that are leading students to drop out or transfer.


This patent takes an active approach to address student retention at universities. Being able to utilize a data-driven approach to identify the students at risk bridges the gap between the students struggling and the systems the university has in place to help retain these types of students. By enabling a machine learning model to be deployed at any university, and from the data available at each distinct university has it developed a specifically catered system to provide the retention system to any school. With this technology, administrators are able to provide better education to their students by determining why their students are more likely to drop out or transfer. they will be able to refine their institution to provide a better education for their students. Written by John DeStefano, Technical Advisor

July 14, 2021