Increasing efficiencies and output through data prediction

Increasing efficiencies and output through data prediction

Getting meaningful information without the hassle

The Customer Challenge

Our customer, a university based in QLD, Admission Assessment team, receives several thousand online international student applications per year, with an influx of applications towards the end of the year. The team’s workload increases dramatically during this period which negatively influences the time it takes to deliver offers to applicants.

The Admission Assessment team manually review applications to identify and prioritise candidates. The overall process was time consuming and as a result, the university believed they were losing strong candidates to other institutions.

To reduce the delay between application submission and initial offer to the applicant, the Admission Assessment team needed a decision support tool that could automatically identify candidates likely to matriculate into their applied for program, secondly, those that are good candidates but whose offer needs to be actioned quickly to avoid the risk of losing them, and thirdly, those that are unlikely to enroll and undertake their chosen program. 

The Solution

TechConnect worked closely with the Admission Assessment team to understand how applications were reviewed and prioritized, and what information was critical to the assessment. 

The Application Ranking solution automatically reviews applications and ranks them according to priority. The solution, collects candidate application data into a machine learning tool that calculates a “confidence score” for each international student’s application which then indicates the likelihood of the student matriculating into their chosen program, based on the content of their application. The confidence score is then used to rank applications to make it quick and easy for the Admission Assessment team to identify promising and time-sensitive applications, with the intent of engaging candidates that are likely to enroll and improve the quality of students accepted into the University’s programs.

Using historical data as a benchmark, the Application Ranking tool provides a reliable 90% accuracy on predicting if a particular student will matriculate based on their application.

The Application Ranking solution is integrated into the University’s Application Management system, enabling the Admission Assessment team to view and manage pre-ranked applications in one place. The automated system has reduced the manual,  time-intensive evaluation and prioritising of applications and increased efficiencies by allowing the team to process applications faster, effortlessly identify priority applicants, and provide offers to candidates quicker, resulting in a higher rate of quality students being enrolled at the university. 

Technology

Amazon S3

 

AWS Lambda

 

AWS Glue

AWS Glue

 

Amazon Athena

Amazon Athena

 

Amazon SageMaker

 

Python