This presentation is prepared as a brief summary for my final thesis viva in the University of Southampton. It also contains slides about self-evaluation of my PhD journey

1. +Viva for the PhD
thesis titled: Prediction
of Course Completion
based on Participants’
Social Engagement on
a Social-Constructivist
MOOC Platform
Ayse Saliha Sunar
28 June 2017
Examiners:
Dr. Dave Millard
Professor Eileen Scanlon
Supervisors:
Dr. Su White
Professor Hugh C. Davis

2. +
Educational Background
 Bachelor: Mathematics at Gazi University, Turkey
 Master: Information Science at Nagoya University, Japan
 Research theme: Intelligent Tutoring System
 Adaptive content presentation based on the level of learners
 Small-scale system
MAGIC
input:
TEXT
output:
MULTIPLE-
CHOICE
CLOZE
QUESTIONS
MAGIC
input:
TEXT
output: ADAPTIVE
PRESENTATION OF
MULTIPLE-CHOICE
CLOZE QUESTIONS
Adaptive
Model
Available system My contribution

3. +
Researching on MOOCs
 PhD: Learning analytics in large-scale data
MOOCs
Pedagogy
Assessments
Stakeholders
Social
learning
through
social
features
Learning
materials
Feedback

4. +
Researching on MOOCs II
 First intention: Developing a friend recommendation system
based on participants’ social engagement
 Cancelled because FutureLearn did not allow us to use an external
tool due to security concerns
 Research project revised: Investigating participants’ social
engagement with the course and predicting course completion
based on social engagement
Hypothesis: The data extracted from participants’ engagement in a
MOOC can be used to identify social behaviour patterns of participants
and this information can contribute to a model of course completion.

5. +
Methodological Phases in my PhD
Descriptive
Statistical
Analysis
• Participants’ contribution to online discussions
• Use of the follow feature, which is one of the unique social feature on the platform
• Relation between learners’ social presence and their performance on course
completion
Inferential
Statistical
Analysis
• The sequence of social behaviours of participants, which are defined with the aid of
characterised use of social affordances on the platform
• Correlation analyses on course completion and behaviour chains
Testing
Prediction
Models
• Prediction of MOOC participants’ course completion performance by using the
pattern of their social engagement in the course
• Random Forest Model and Support Vector Machine

6. +
Contributions of my PhD
 MOOC learners who participated socially are more likely to
complete the course than others.
 Some social features on the platform are actually a good
indicator for learners’ patterns of engagement.
 Sequence of learners’ social behaviours can be modelled as
behaviour chains.
 This social behaviour modelling is valuable since it identifies
which components of social behaviours might be the best
discriminators for predicting learners’ future behaviours.
 A prediction model of course completion has been developed
by using behaviour chains.
 The model performed well especially in classifying learners as
“to be in low or high completion”.

7. +
--- End of the brief summary ---

8. +
Best things I have learnt
 How to individually conduct studies
 How to effectively use resources on a particular study
 How to gain my academic networks
 How to constructively criticise other research
 Time and task management
 Academic writing

9. +
Difficulties during the study
 Lack of some data
 No data regarding to like on the platform
 More accurately gathered follow data
 Issue on FutureLearn’s authorisation on third party
involvements on the platform
 Was informed one week before the transfer viva (at the 24th month
of the PhD)
 Had to do major changes in the study
 Machine Learning
 Hard to learn in the beginning compare to any other new subjects
that I have learnt during the study

10. +
After the PhD
 Soon:
 More publication
 Behaviour chains and prediction models
 Soon:
 Continue to collaboration with Yildiz Technical University on second
language MOOC participants
 In 6 months:
 Start the work in a university in Turkey
 As a lecturer at the department of
Computer Engineering
 Seek for new collaboration opportunities on MOOCS in the context of
education in Turkey
Karabuk Bitlis

11. +
What would I have changed if I had
a chance to do this PhD again?
 Do more brain storming with other fellow learners
 I did it when I really struggled with a problem.
 If I had more data on learners’ social interactions on the
platform such as likes and if learners actually read the
comments, it could provide insight to learners behaviours.