Hellenic Open University
A holistic approach to learning analytics and educational data mining in distance learning through data and machine learning pipelines
Abstract
dc:descriptionThe transition of our civilization to the online technologies and the metaverse, generated huge data sources and elaborated methods of processing. By mimicking the human brain, organizations of every field are trying to establish processes of high complexity to extract knowledge from the available data. At the same time, the changes in the lifespan of the validity of information creates the necessity of continuous education. This results in a series of changes and challenges in teaching and learning, mainly in the landscape of higher education. Thus, educational organizations need to take action to leverage data for evidence-based decision making.Learning Analytics (LA) and Educational Data Mining (EDM) tools are specifically developed to manage educational data by predicting patterns and behaviors. They integrate data related to students, instructors, classes and courses, grades, forums etc. for processing, monitoring and visualizing. The ultimate goal of this kind of applications is the improvement of the learning environment for the benefit of the stakeholders. However, LA and EDM tasks are often challenging, involve the use of various tools and demand serious amounts of time and effort to repeat them. In order to acquire the necessary information for revolutionizing learning, small scale LA projects are not enough. It is required to adopt a holistic approach where end-to-end solutions would begin from data gathering and would result in reporting actionable knowledge. In the last years there is a rising trend in the use of data pipelines. Pipelines are collections of processes that can transfer and transform data from various sources, aiming to create new knowledge. They come with several benefits that can also be applied in a LA cycle.This thesis consists of two main parts. The first part proposes a LA framework and the respective experiments that aim to provide a better understanding on students’ learning in a Distance Education environment. The experiments shift focus from simple log data to a combination of primitive and derived variables taking into account the social aspect of learning. Using a variety of methods, including simple visualizations and elaborated analytical processes, students’ latent traits can be revealed, offering knowledge that can guide decisions for teaching and learning improvement The second part introduces the use of data pipelines for performing LA processes. In several projects, data and machine learning pipelines were designed, created and executed. It was shown that the pipelines can undertake the whole LA cycle and also produce interactive LA dashboards. By approaching LA and EDM as a holistic, repeatable and reliable process in a multidisciplinary synergy, important benefits can emerge for the teaching and learning process.
Degree
thesis:*- Grantor dc:publisher
- Hellenic Open University
- Year dc:date
- 2022
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Tsoni, Rodanthi
- Τσώνη, Ροδάνθη
Subjects
dc:subject × 20- Αναλυτική της μάθησης
- Εξόρυξη Εκπαιδευτικών Δεδομένων
- Εξ αποστάσεως εκπαίδευσης
- Αγωγοί Δεδομένων και Μηχανικής Μάθησης
- Learning analytics
- Educational data mining
- Distance learning
- Data and Machine Learning Pipelines
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Φυσικές Επιστήμες
- Τεχνητή νοημοσύνη
- Εκπαίδευση
- Κοινωνικές Επιστήμες
- Εκπαίδευση και Έρευνα της εκπαίδευσης
- Computer and Information Sciences
- Natural Sciences
- Artificial Intelligence
- Education
- Social Sciences
- Education and Educational Research
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/53168