Publications
De Freitas, K., & Bernard, M. (2017). Enabling Integrated Data Mining Analysis in Moodle with FlexEDM, In Proceedings of the International Conference on e-Learning, 1-2 June 2017, The University of Central Florida, Orlando, USA.
De Freitas, K., & Bernard, M. (2015). Comparative Performance Analysis of Clustering Techniques in Educational Data Mining. IADIS International Journal on Computer Science & Information Systems, 10(2), pp. 65-78.
De Freitas, K., & Bernard, M. (2015). Comparative Analysis Of Clustering Algorithms Within A Web-Based LMS, 9th European Conference on Data Mining (ECDM’15), 22 – 24 July 2015, Las Palmas de Gran Canaria, Spain, pp.
De Freitas, K., Bernard, M., (2014). A Framework for Flexible Educational Data Mining, Proceedings of the 2014 International Conference on Data Mining, DMIN2014, WorldComp 2014, Las Vegas, USA, pp. 176–180.
Research Areas
Exploring the Creation of a Graph-Based Database Model for the Moodle
Description: Utilised Moodle LMS and Neo4J Graph-DB to represent and analyse user-interactions for predicting goals. Compared relational and graph-based models across several criteria.
Knowledge Area: Educational Data Mining
Student: Willis Estrada (Completed 2017)
| Neo4J | |
| Moodle | |
| PHP |
Automated Generation of Mobile and Cloud Systems using Domain Models
Description: Utilized a JSON-based model to build front-end application in React-Native and backend CRUD operations in Google Firebase Cloud Service.
Knowledge Area: Mobile Learning, Software Engineering
Student: Dillon Bhola (Completed 2018)
| React-Native | |
| Google Firebase Database |
Using Crowd-sourced Location Data to Generate Real Time Transit Estimates
Description: Utilised REST-based services to crowdsource transit information from Android devices for modelling transit times and suggest alternatives.
Knowledge Area: Mobile Learning
Student: Javed Ali (Completed 2018)
| Android (Java) | |
| PHP (Slim) | |
| MySQL |
Offline Handwriting Recognition on Mobile Devices using DNN
Description: Developed a distributed strategy to update TensorFlow models utilised and captured on Android devices.
Knowledge Area: Mobile Learning, Prediction with ML
Student: Zane Oliver (Completed 2017)
| TensorFlow | |
| Android (Native) |
Data Driven Approach for Fake New Detection using Big Data Fusion Techniques
Description: Utilised the JDM data fusion model to merge sentiment analysis, text classification, and structural analysis to predict fake news articles.
Knowledge Area: Web Mining
Students: Kemar Celestin (Completed 2017)
| IBM Watson |
Using DNN to Improve Country Identification of Speech
Description: Built database of Caribbean voices. Built a Deep Neural Network model that recognises variations in speech patterns to identify country of origin.
Knowledge Area: Prediction with ML
Students: Kerschel James
| TensorFlow | |
| NodeJS | |
| Web-API (voice recording) |
Benchmarking and Implementation of Popular Machine Learning Models for Churn Prediction in Telecommunications
Description: Implemented and benchmarked popular machine learning models using a dataset from a Caribbean-based telecommunication company. Additionally, gradient boosted trees, was also implemented and performed most accurately compared to the other models.
Knowledge Area: Prediction with ML
Student: Randy Ram (Completed 2018)
| XGBoost | |
| SciKit-Learn | |
| Oracle |