An integrated framework for learning analytics

Soltanpoor, R 2018, An integrated framework for learning analytics, Doctor of Philosophy (PhD), Science, RMIT University.


Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Soltanpoor.pdf Thesis application/pdf 23.82MB
Title An integrated framework for learning analytics
Author(s) Soltanpoor, R
Year 2018
Abstract Low retention rates have been an ongoing concern, especially among educational institutions amidst expanding their student base and catering to large and diverse student cohorts. Increasing retention rates without lowering academic standards poses many challenges. The traditional teaching techniques using a one-size-fits-all approach appear to be less effective, and the size and diversity of cohorts demand innovative teaching techniques allowing for adaptive and personalized coaching and learning.

In this thesis, we propose a novel, adaptive and integrated analytics framework for learning analytics to address the key concerns of educational institutions. The proposed framework comprises three layers: (1) the conceptual layer which is a context-agnostic and generic analytics layer including descriptive, predictive, and prescriptive techniques; (2) the logical layer or the context-specific learning analytics processes layer that specializes the conceptual layer in the context of education; ten key learning analytics processes are formalized, implemented, and linked to the conceptual layer components; finally, (3) the physical layer that is concerned with education-oriented application implementations and is a context-specific components/algorithmic implementation of the logical layer processes. Our proposed framework, however, is not limited only to the learning and teaching environment. As a proof of concept, we chose the education context and applied our framework on it. The three-layered integrated learning analytics framework proposed allows domain-agnostic elements defined in the conceptual layer to be realized by domain-specific processes in the logical layer, and implemented through existing and new components in the physical layer. Please note that the learning analytics is not confined to the education context alone. The framework, therefore, can be customized for different domains making the approach more widely applicable.

An adaptive and innovative approach in the physical layer named the personalized prescriptive quiz (PPQ) is introduced as a demonstration of education-oriented applications assisting the educational institutions. The novel agile learning approach proposed combines descriptive, predictive and prescriptive analytics to create a personalized iterative and incremental approach to learning. The PPQ allows students to easily analyze their current problems (especially, identifying their misconceptions), predict future results, and benefit from personalized intervention tasks. The enhanced PPQ incorporating difficulty and discrimination indexes, run-time question selection, and a hybrid iterative predictive model can be more beneficial and effective for personalized learning.

The results demonstrate a significant improvement in student academic performance after applying the PPQ approach. In addition, students claimed that the PPQ helped them elevate their self-esteem and improve student experience which may eventually lead to improved retention rates.
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Science
Subjects Database Management
Keyword(s) Learning Analytics
Adaptive Learning
Personalized Learning
Prescriptive Analytics
Analytics Framework
Versions
Version Filter Type
Access Statistics: 94 Abstract Views, 162 File Downloads  -  Detailed Statistics
Created: Wed, 20 Mar 2019, 15:27:21 EST by Adam Rivett
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us