From Analytics to Cognition: Expanding the Reach of Data in
Learning
R. Tsoni
1
, C. Samaras
1
, E. Paxinou
1
, C. Panagiotakopoulos
2
and V. S. Verykios
1,a
1
School of Science and Technology, Hellenic Open University, Patra, Greece
2
Department of Primary Education, University of Patras, Patra, Greece
Keywords: Data Mining, Learning Analytics, Hellenic Open University, Big Data Lab.
Abstract: Education constitutes a rapidly changing and challenging environment, therefore, a shift from reporting to
actionable interventions based on data is almost imperative. At the same time, Learning Analytics is
incorporating increasingly advanced tools and methods. Artificial Intelligence and cognitive science allow us
to study in depth the behavior of students, creating patterns and prediction models. All the above summarize
the scope of the newly established Big Data Analytics and Anonymization Laboratory (BAT Lab) in the
Hellenic Open University. This paper presents the vision and the work in progress of the BAT Lab in an
attempt not only to produce interpretable results but also to organize and present these results in a highly
usable way for non-experts. Additionally, PRIME-EDU software, which is the teams latest work, is presented.
1 INTRODUCTION
1.1 The Era of Data Analytics
Higher Education Institutions are facing major
difficulties which will, at some point, render them
incapable of sustaining their existence. Some of these
problems include outdated curriculum, high rates of
students’ dropouts, and graduates who lack the
necessary skills for the modern job market. This leads
to unqualified workforce and outdated and irrelevant
institutions (DeMillo, and Young, 2015). It is obvious
that changes need to be made.
Contemporary methods to data-driven decision
making include controlled experiments like A/B
testing and the use of questionnaires. However, all
these methods present certain problems. Regarding
experiments, scalability is a big issue, and as far as
questionnaires are concerned, they are not considered
reliable as people’s answers are not always
representative of their true feelings (Lindstrom,
2010). Manual approaches to gathering data incur a
very high cost as well.
It appears that there must be a more suitable
alternative. Living in the Petabyte age equals rapidly
developing technology and more information
gathering than ever before. This has led to the
a
Work by Vassilios S. Verykios was partially supported by a research grant from New York University Abu Dhabi.
creation of Big Data research area which hopefully
will provide the answers to some of the problems that
society and academia are facing today. In order to do
this, data mining and data analytics are employed, by
using a well-established iterative process for model
building and evaluation. It used to be the case that the
focus of this process was on the practical steps,
whereas it has now become apparent that the pressing
need is to be able to tell a story with the data (Knaflic,
2015). A story that aims to shed light on the issues
raised in education.
1.2 Learning Analytics: An Attempt
for an Efficient and Innovative
Model for Educational
Development
Education, as a living and rapidly transforming
organism, needs a sophisticated method to collect,
analyze and act upon data. Actually, higher levels of
complexity demand more sophisticated ways of
information processing. Business has benefited and
ripped rewards for the insights provided by data
analytics and it would not be an exaggeration to say
that is has revolutionized the way commerce works
today. It would be really useful if education
458
Tsoni, R., Samaras, C., Paxinou, E., Panagiotakopoulos, C. and Verykios, V.
From Analytics to Cognition: Expanding the Reach of Data in Learning.
DOI: 10.5220/0007751904580465
In Proceedings of the 11th International Conference on Computer Supported Education (CSEDU 2019), pages 458-465
ISBN: 978-989-758-367-4
Copyright
c
2019 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
stakeholders could find ways, consistent with
pedagogical values and related to the educational
demands, to apply these successful models with the
same extraordinary results. That is all about Learning
Analytics.
Learning Analytics involves analyzing data that has
been collected from different educational contexts and
environments, and from various levels with the view of
discovering patterns. By gaining this knowledge,
decisions can be made and changes can be
implemented which will bring about favorable
outcomes for the students and the institutions.
Hellenic Open University (HOU) is the only Open
University in Greece. It offers exclusively distance and
blended learning courses to mature students of diverse
backgrounds, skills, and qualifications. The unique
nature of this institution means that it confronts a
unique set of problems as far as meeting the needs of
its students is concerned. Finding ways to
accommodate such a diverse population and optimize
the learning experience is certainly a challenge that led
to the creation of the Big Data Analytics and
Anonymization Laboratory (BAT lab).
2 BAT LAB PRESENTATION
2.1 The Objective
BAT lab conducts research in the field of large-scale
data management and analysis, in conjunction with
privacy protection of this data with the view to
understand student behavior and interaction both with
their peers but also with the teaching staff. Its main
purpose is to find ways to delve into the hidden patterns
relating to student behavior, communication,
engagement, performance and faculty guidance.
2.2 Difficulties and Barriers
Both children and adolescents are familiar with the
idea of finding and using digital information. For them,
data is ubiquitous and they pursue the creating, using
and sharing of data. They own smartwatches to
monitor their physical activity, they use social media to
state their emotions and they measure their popularity
by counting likes and views. Unlike the new
generation, a lot of academic members are still using
conventional methods to get feedback from their
students and theories to predict the efficiency of the
educational material they use. Thus, a significant
barrier is posed by a pre-existing culture or, in some
cases, even a sense of technophobia. So, in order to
obtain solutions, it is imperative that the data is
analyzed in a way that is intuitive and at the same time
more user-friendly, to reveal hidden models or patterns
for prediction and summarization, which are not
initially obvious (Siegel, 2013). A shift from metrics to
analytics and from reporting to actionable
interventions is the next generation of the learning
environment. However, this transition will require a
significant institutional change (Baer, and Campbell,
2012).
Another, more practical, barrier is the fact that almost
all the available NLP (Natural Language Processing)
tools do not support Greek language. Specific tools for
sentiments analysis have been created by Greek
researchers (Agathangelou et al., 2018) and are used by
the team for analyzing students’ fora (see section 3.4).
3 PROJECTS AND ONGOING
RESEARCH
3.1 Outlining the Identity of HOU
HOU is University unique at its scope in Greece, as it
serves the ideal of openness and inclusiveness. This
openness leads to a big diversity amongst the students
who choose to be educated through this system. So, it
was crucial to be able to gather and analyze
successfully all this information that comes from the
students’ admissions.
For example, data coming from HOU students’
admission in the last four years analyzed in BAT Lab
showed that there is a greater number of female
applicants, although there is a progressing decline in
the total number of applicants throughout the years
(figure 1). School of Humanities is much more popular
among the female applicants, whereas in the School of
Science and Technology the male applicants far
Figure 1: Number of applications per gender during 2003-
2013 in HOU (Stavropoulos et al., 2017).
From Analytics to Cognition: Expanding the Reach of Data in Learning
459
outnumber the female. According to figure 2, the
highest percentage of the applicants comes from the
regions of Attiki, Achaia, and Thessaloniki although,
for the rest of the country, there is not a particular
pattern that can be detected.
Figure 2: Applications per region (Verykios &
Stavropoulos, 2018).
3.2 Assisting Tutors to Catch up with
Students’ Progress
The percentage of the students who submitted
assignments and quizzes, or the level of the students’
and the tutors’ engagement, or other relevant
information is very significant for online courses
where the transactional distance is longer than it is in
a face to face course (Moore, 2013). For example,
data on how long does a student stay connected to the
platform or on how many connections does she/he
make to the platform on average per day, provides
information on student’s dedication time. The same
information is also available about tutors (figure 3). It
is notable that even though some tutors seem to
dedicate the least amount of time, in comparison to
their colleagues, they still make an adequate number
of connections. This sort of information allows the
module coordinator to create some guidelines, in a
sense of best practices and benchmarks for the tutors
in order for the latter to become more efficient in their
work and for the coordinator to access more
objectively tutors’ work.
Another piece of information that could help the
tutors become more effective in their roles, is the
students’ engagement in relation to the time of the
day. It appears that students are more active in the
evenings (figure 4) which is something that makes
sense since the majority of the students in HOU are
mature students who have family and job obligations
during the rest of the day. Thus, it would be more
Figure 3: Tutors' connections per day.
Figure 4: The distribution of active students and students’
activities to daily hours (Gkontzis et al., 2017a).
productive for tutors to be also connected to the
university platform in the evening rather than in the
morning when students engagement appears to be
very low. As a result, every student communicates
directly with the tutor, asking for advice and help and
the whole educational process becomes more
productive for both of them. The student progress bar
(figure 5) shows how many of the allocated activities
a student completes. From the overview page, the
tutor can compare every student’s performance to the
performance of the whole class.
Figure 5: Students' progress bar (Gkontzis et al., 2017a).
All these important facts allow tutors to be supportive
to students, provided that these data are easily
accessible. Thus, visualization is a major concern.
Learning Analytics dashboards had been successfully
used, enabling tutors with no previous LA training to
detect students trends in order to provide
personalized assistance (Gkontzis, et al., 2017).
Furthermore, students’ logins, replies, and
quizzes were blended with the average grade of the
main written assignments throughout an academic
CSEDU 2019 - 11th International Conference on Computer Supported Education
460
year. Useful observations from the students’
educational online activities were made that can be
used as predictive factors for their academic
performance (Gkontzis, et al., 2018a). In a related
work Gkontzis, et al. (2018b) presented an analytical
framework that provides a comparison between
students’ actual predicted grade, identifying students
who have included third-party services in their
assignments.
3.3 Evaluating the Efficiency of
Methods and Educational Material
While it is true that big data is a field of immense
interest, unfortunately, the spotlight is on how to
store, index, retrieve and interrogate the data rather
than how to analyze and utilize it in an efficient and
user-friendly way which is a shame as the answers to
most problems are hidden within this ocean of data
(Verykios & Stavropoulos, 2018).
In a project run by Paxinou, et al., (2017) the
authors presented that the Classical Test Theory
(CTT) method is not able to prove successfully the
lead of a virtual reality laboratory in preparing
Science students in HOU for the experiments in the
wet lab, as this method doesn’t separate the difficulty
of the questions in the written tests from the students’
ability. Contrariwise, the applications of more
advanced theories and methods, as the Item Response
Theory (IRT) allows highlighting the positive impact
of a teaching method to the students’ learning
outcomes. This result poses a significant issue of the
importance of choosing a suitable analysis framework
in order to achieve a result that best describe reality.
There is a great value in distinguishing the skill level
of a student from the difficulty level of the questions
of each evaluation test. This significant result that
takes advantage of the IRT model, was proven from a
massive data analysis of a several million recorded
online games of chess held by Anderson from
Microsoft, Kleinberg from Cornell University and
Mullainathan from Harvard University (2016). Their
research brought to light unquestionable proof for the
players’ behavior concerning decision making. The
same way Guo et al., (2014) changed the way that
educational videos are made by analyzing a vast
number of MOOC data concerning students’
engagement.
An interesting study highlighted the importance
of data analysis for profiling new coming university
students. The study was conducted through four
successive years in freshmen students of the School
of Education at Patras University in a collaborative
project between BAT Lab and University of Patras
(Panagiotakopoulos et al., 2017). Data analysis
revealed the difference between studentsopinion and
their actual attitudes. Although they stated their
acknowledgment for the significance of technology
use for educational purposes, data showed that they
barely take advantage of technology in their studies.
Concerning HOU students, several sets of data have
been analyzed in order to find out the level of
engagement with the educational material. The
graphs in figure 6 illustrate the engagement students
have with the content provided for them on their
course. A large number of students who do not use
the resources indicates that perhaps the content is
problematic and needs to be rethought by the tutors
and coordinator.
Figure 6: Students' access to content material.
3.4 Gaining Insight into Students’
Behavior
In an effort to see beyond logins and submission
dates, network analysis was conducted. These graphs
(figure 7) show the interaction between students and
tutors, and between the students themselves. This
gives an overview of the communication that is taking
place and it allows us to see how a tutor is interacting
with his/her group.
Sentiment analysis of students’ fora and emotion
classification (figure 8) was held by Gkontzis et al.,
(2017a) in postgraduate students of the School of
Science and Technology in HOU. The study revealed
that positive polarity was dominant in students’ posts.
This fact was more obvious in the most active
students, indicating an emotion of satisfaction.
Additionally, this work pointed out some issues for
future research: possible correlation of sentiment with
From Analytics to Cognition: Expanding the Reach of Data in Learning
461
Figure 7: An integrated view of the groups’ forum graph.
academic and demographic factors, more active
students’ sentiment compared with less active ones
and the effect of tutors methods in students’
sentiment expressed in fora.
Figure 8: Emotion classification in forum posts (Gkontzis,
et al., 2017b).
Text mining in forum data whereby individual terms
are picked out of texts sent by students has been used
in an attempt to better understand their educational
needs. A high frequency indicates that students are
having difficulties due to misconceptions or require
further clarification. The tutors can intervene by
providing more information on the subject or
improving their teaching of the subject.
3.5 The Prime-EDU Software
Recently our team has designed and developed a new
educational application named PRIME-EDU. Its
purpose is to transfer educational data to a Cloud
Database. These data are analyzed using Learning
Analytics techniques so that the application users can
easily make better educational decisions. The above-
mentioned Data Base is designed in the form of Cloud
Figure 9: Data visualisation in PRIME-EDU software.
Warehouse Database and its data provide real-time
Visualization, Correlation, and Prediction.
The PRIME-EDU application is designed to receive
data from “MySchool”, “Moodle”, the “DIARROI”
(STUDENTS DROP OUT) application and the
“Classter” of Vertitech. “My School” is the web-
based application of the Ministry of Education, that
all schools of Secondary Education in Greece use
compulsory. Moodle is one of the most established
asynchronous e-learning systems that offers many
opportunities for analyzing training data. It was also
used by HOU. Finally, the DIARROΙ application is a
training software that our team has developed which
offers telematics services to reduce early school drop-
out (ESL) and educational leakage (Samaras et al.,
2018).
The PRIME-EDU application is designed with C# in
Microsoft Visual Studio 2017. It makes use of the
Microsoft Virtual Machine libraries. NET
Framework 3.5 works on various versions of
Windows.
Figure 10: PRIME-EDU Interface.
The application automatically sends learning data to
the Warehouse Database from Excel, Access, SQL
SERVER. It is also texting information to parents,
such as updating for their children's attendance. The
basic requirement of the application is that the data
sent to the Cloud Warehouse Database must be
anonymous in order to comply with the European
Data Protection Regulation. Besides that, personal
data don’t help us make educational decisions.
CSEDU 2019 - 11th International Conference on Computer Supported Education
462
The PRIME-EDU application transfers data to the
Cloud Warehouse Database, that has been developed
on SQL Server 2017. The Warehouse Database is
designed with a table of events and dimensional
tables in a star model. The event table has each
student's daily attendance and the dimensional tables
contain the most features that affect the student's
attendance such as environment, school teacher,
family etc.
From the Cloud Warehouse Database, we isolate the
data needed to make educational decisions. The
technique we use is the business intelligence that
companies with a lot of data have been using for many
years in order to monitor the behavior of their
customers and their organization. Thus, with CUDE
and ROLLUS techniques, we are able to isolate an
educational phenomenon and handle it per period, per
region, per studentsage etc. With the data that we
isolate with CUBE techniques and the abilities
(potentials) of the R Tool Visual Studio (RTVS) tool,
we are able more easily visualize an educational
phenomenon, to apply Correlation and look for the
forecast.
3.6 Privacy and Anonymization
A significant field of concern is privacy protection
and data anonymization. This concern also applies to
the educational field. In order to protect students that
provide data, it is necessary to omit those
characteristics that permit identification. Relevant
methods and techniques are presented in a series of
successive works as the followings: Sakkopoulos et
al., (2013) presented a framework that allows
conducting research on anonymity techniques in a
real-life environment using smartphones. The
framework also includes logging mechanisms that
facilitate positioning research dataset development in
open format. Additionally, Karapiperis, & Verykios,
(2016) proposed an efficient scheme for privacy-
preserving record linkage by using the Hamming
locality-sensitive hashing technique as the blocking
mechanism and the Bloom filter-based encoding
method for anonymizing the data sets at hand. Highly
accurate results were achieved and simultaneously
reduced significantly the computational cost by
minimizing the number of distance computations
performed. Moreover, Bit Vectors (BV), an accurate
distance-preserving encoding scheme for
representing numerical data values in privacy-
preserving tasks, were used by Karapiperis, et al., in
their work (2017). Key components of this
embedding process were the employed hash functions
and the threshold that is required by the distance
computations, which they proved that can be
specified in a way that guarantees accurate results.
Finally, it is noted that as the level of anonymization
rises, the quality of data declines. Consequently, it is
important to achieve a balance between tanking
advance of data the best possible way and complying
with privacy regulations.
4 CONCLUSIONS AND FUTURE
WORK
4.1 A Holistic Approach
Overall, it is obvious that Learning Analytics and its
application to different contexts and environments
has the power to transform the educational system as
a whole, as well as to tailor the learning experience to
the specific needs of individual students which will
lead eventually to their success. Using the resources
available in ΒΑΤ lab, we have been able to conduct
research and experiment with the implementation of
various Learning Analytics case studies with some
promising outcomes. Even though we are at an initial
stage we hope to be able to discover ways to
revolutionize our learning and teaching environments
by offering our students the best possible services.
4.2 From Learning Analytics to
Cognitive Science
Since translating experience into words makes a
major difference in future actions (Pennebaker et al.,
2014), analyzing written text is meaningful in
education as it can reveal insights that go deeper than
grades or login counts. Therefore, the use of Natural
Language Processing methods is crucial in an
integrated educational data analytics approach.
Cognitive computing combines neuroscience,
supercomputing, and nanotechnology to develop a
coherent, integrated, universal mechanism inspired
by the mind's capabilities Modha et al., (2011).
Learning Analytics combined with State-of-the-art
Cognitive Computing can provide new effective ways
for content management, especially in Distance
Learning Universities where the educational material
has to fill the gap in the lack of tutor’s physical
presence. An example is the work of Des et al.,
(2018) where they used the aforementioned methods
to manage the content of micro-learning videos in
order to improve their retrieval from students. By
using sophisticated systems that provide assistance to
From Analytics to Cognition: Expanding the Reach of Data in Learning
463
students, at the same time a pool of data is created,
allowing the gaining of insight into their behavior and
their way of learning. Cognos Analytics provides a set
of resources for data analysis that can uncover hidden
patterns that could lead to significant conclusions
even in cases when the initial hypothesis is not clearly
stated. This kind of serendipity is what we consider
as the main benefit of the employment of
continuously integrated methods and tools.
4.3 The Future Scope
In the future, we hope to be able to create increasingly
sophisticated applications and gadgets and to create a
kind of augmented reality environment which can
communicate in real time how the student is
performing. By being able to visualize their progress
both the students and the faculty will be able to avoid
negative outcomes, such as poor grades or even worse
students’ dropouts. In a recent study, Brown et al.,
(2019) provided evidence that large scale data form
application users via smartphones can outweigh the
noise inherent in collecting data outside a controlled
laboratory setting and produce valid results. This
conclusion supports the idea of the creation of an
online educational application that would gather and
present to students selected data in an understandable
and useful way in order to enhance their self-
regulation and their awareness about their progress.
The goal is to make optimal use of all available data.
After all, as Schmidt (2011) stated: Technology is
not really about hardware and software anymore. It’s
really about the mining and use of this enormous
volume of data in order to make the world a better
place…”.
ACKNOWLEDGMENTS
The co-authors would like to acknowledge the
support of New York University Abu Dhabi.
REFERENCES
Agathangelou, P., Katakis, I., Koutoulakis, I., Kokkoras, F.,
and Gunopulos, D. (2018). Learning patterns for
discovering domain-oriented opinion words.
Knowledge and Information Systems, 55(1), 45-77.
Anderson, A., Kleinberg, J., and Mullainathan, S. (2017).
Assessing human error against a benchmark of
perfection. ACM Transactions on Knowledge
Discovery from Data (TKDD), 11(4), 45.
Anderson, C. (2008). The end of theory: the data deluge
makes the scientific method obsolete. Wired Magazine
16.07. Retrieved by:
https://www.wired.com/2008/06/pb-theory/
Baer, L., and Campbell, J. (2012). From metrics to
analytics, reporting to action: Analytics’ role in
changing the learning environment. Game Changers:
Education and Information Technologies, Educause,
53-65.
Brown, H. R., Zeidman, P., Smittenaar, P., Adams, R. A.,
McNab, F., Rutledge, R. B., and Dolan, R. J. (2014).
Crowdsourcing for cognitive sciencethe utility of
smartphones. PloS one, 9(7), e100662.
DeMillo, R. A., and Young, A. J. (2015). Revolution in
higher education: How a small band of innovators will
make college accessible and affordable. MIT Press.
Dessì, D., Fenu, G., Marras, M., and Recupero, D. R.
(2018). Bridging learning analytics and Cognitive
Computing for Big Data classification in micro-
learning video collections. Computers in Human
Behavior.
Gkontzis, A. F., Karachristos, C. V., Lazarinis, F.,
Stavropoulos, E. C., and Verykios, V. S. (2017a).
Assessing Student Performance by Learning Analytics
Dashboards. 9th International Conference in Open and
Distance Learning, 9(1A), 101-115.
Gkontzis, A. F., Karachristos, C. V., Panagiotakopoulos, C.
T., Stavropoulos, E. C., and Verykios, V. S. (2017b).
Sentiment Analysis to Track Emotion and Polarity in
Student Fora. In Proceedings of the 21st Pan-Hellenic
Conference on Informatics (p. 39). ACM.
Gkontzis, A., Kotsiantis, Panagiotakopoulos, C.,
&Verykios V. (2018a). Measuring Engagement to
Assess Performance of Students in Distance Learning.
Proceedings of the 9th International Conference on
Information, Intelligence, Systems and Applications,
Zakynthos, Greece. IEEE
Gkontzis, A., Kotsiantis, S., Tsoni, R., and Verykios, V.
(2018b). An Effective LA Approach to Predict Student
Achievement. In Proceedings of the 22nd Pan-Hellenic
Conference on Informatics. ACM.
Guo, P. J., Kim, J., and Rubin, R. (2014). How video
production affects student engagement: An empirical
study of mooc videos. In Proceedings of the first ACM
conference on Learning@ scale conference (pp. 41-50).
ACM.
Karapiperis, D., and Verykios, V. S. (2016). A fast and
efficient Hamming LSH-based scheme for accurate
linkage. Knowledge and Information Systems, 49(3),
861-884.
Karapiperis, D., Gkoulalas-Divanis, A., and Verykios, V. S.
(2017). Distance-aware encoding of numerical values
for privacy-preserving record linkage. In Data
Engineering (ICDE), 2017 IEEE 33rd International
Conference on (pp. 135-138). IEEE.
Knaflic, C. N. (2015). Storytelling with data: A data
visualization guide for business professionals. John
Wiley and Sons.
CSEDU 2019 - 11th International Conference on Computer Supported Education
464
Knaflic, C. N. (2015). Storytelling with data: A data
visualization guide for business professionals. John
Wiley and Sons.
Lindstrom, M. (2010). Buy ology: Truth and lies about why
we buy. Crown Business.
Modha, D. S., Ananthanarayanan, R., Esser, S. K.,
Ndirango, A., Sherbondy, A. J., and Singh, R. (2011).
Cognitive computing. Communications of the ACM,
54(8), 62-71.
Moore, M. G. (2013). The Theory of Transactional
Distance. In Handbook of distance education (pp. 84-
103). Routledge.
Panagiotakopoulos, C., Koustourakis, G., Samaras, C.,
Stavropoulos, I., and Verykios, V., (2017). The
Transition from High School to University: Means of
Learning and Technology Awareness for Junior
Students - Trainee Teachers. 9th International
Conference in Open and Distance Learning. 9(2Α),
138-153.
Paxinou, E., Sgourou, A., Panagiotakopoulos, C., and
Verykios, V. (2017). The item response theory for the
assessment of users’ performance in a biology virtual
laboratory, Open Education: the journal for Open and
Distance Education and Educational Technology,
13(2),107-123.
doi:http://dx.doi.org/10.12681/jode.14618
Pennebaker, J. W., Chung, C. K., Frazee, J., Lavergne, G.
M., and Beaver, D. I. (2014). When small words foretell
academic success: The case of college
Sakkopoulos, E., Mersini, P., Tsakalidis, A., Sioutas, S.,
and Verykios, V. (2013). A Novel Mobile Framework
for Anonymity Techniques and Services Research. In
Mobile Data Management (MDM), 2013 IEEE 14th
International Conference on (Vol. 1, pp. 353-355).
IEEE.
Samaras, C., Panagiotakopoulos, C., and Verykios, V.
(2018). Διερεύνηση της φοίτησης με Learning
Analytics: Μελέτη της ετήσιας φοίτησης των μαθητών
του ΕΠΑ.Λ. Δοξάτου 2016-2017. 11th Pan-Hellenic &
International Conf. ICT in Education
Siegel, E. (2013). Predictive analytics: The power to predict
who will click, buy, lie, or die (p. 148). Hoboken:
Wiley.
Stravropoulos, E. C., Panagiotakopoulos, G. T., Kagklis,
V., Lionarakis, A., Marketos, G., and Verykios, V. S.
(2017). Student Admission Data Analytics for Open
and Distance Education in Greece. The Journal for
Open and Distance Education and Educational
Technology, 13(2), 6-16.
Verykios, V. S., and Stavropoulos, E. C. (2018). Exploring
the Power of Learning Analytics. The Envisioning
Report for Empowering Universities, 9.
From Analytics to Cognition: Expanding the Reach of Data in Learning
465