You are here

Predicting Students Success with Leganto, A Proof Of Concept Machine Learning Project

Speakers
Mr Peter Green
1, Mr Gal Darom2, Mr Tomar Katz2, Mr David Lewis1

1Curtin University, 2Ex Libris

Abstract
Predicting Students Success with Leganto, a proof of concept machine learning project.

Curtin Library successfully implemented the Ex Libris Leganto so ware in 2016 to provide an integrated, client facing Reading List solution. The implementation of Leganto was well received by instructors and has become the common way in which students discover and access their Reading Lists at Curtin University.

Ex Libris approached Curtin Library in 2017 with a proposal for a proof of concept project which would investigate the correlation between student success and activity within the Leganto Reading List. It is hoped that this will enable early identification of students who are likely to struggle and will provide an opportunity for early intervention with strategies to mitigate the risk of failing. Using learning analytics to predict student success and to identify students at risk of failing to complete their studies has been an ongoing area of interest to Curtin University and this o er from Ex Libris to partner in a proof of concept project was accepted.

To identify student activities that might be associated with learning success Ex Libris would analyse Leganto usage data combined with extra data provided by Curtin Library to identify characteristics of the courses, student course results and basic student profile information. Providing this extra data was an essential part of the project and one that required a clear understanding between Curtin Library and Ex Libris about the way in which the data would be handled to ensure that student privacy would be ensured.

During 2018 Ex Libris applied data mining and machine learning algorithms to discover insights in the data, building prediction models to assist in identifying students who are more likely to struggle. Early outcomes indicate that predictive models could be developed. With further development these models might then be applied in real time, allowing the institution to focus its retention efforts on students who fall in the high risk category.  Following the successful proof of concept project this functionality might be integrated with Leganto or added as a service offered to Leganto customers.