Guest editorial
| Date | 20 May 2019 |
| Published date | 20 May 2019 |
| DOI | https://doi.org/10.1108/IDD-05-2019-074 |
| Pages | 65-66 |
| Author | Xu Du,Jui-Long Hung,Chih-Hsiung Tu |
| Subject Matter | Library & information science |
Guest editorial
Enhancement of teaching and learning:
applications in learning analytics and educational
data mining
Learning analytics (LA) and educational data mining
(EDM) are highly related subjects that overlap in definition
and scope. Although both communities of researchers
within LA and EDM have similarities where learning science
and analytic techniques intersect, there are some significant
differences between them in terms of origins, techniques,
fields of emphasis and types of discovery (Siemens and
Baker, 2012;Chatti et al., 2012;Romero and Ventura,
2013). EDM refers to computerized methods and tools for
automatically detecting and extracting meaningful patterns
and information from large collections of data from
educational settings (Kumar and Sharma, 2017). LA is
focused on understanding and optimizing learning and
learning environments by measuring, gathering, analyzing
and reporting of data about learners and learning contexts
(Siemens and Baker, 2012).
The aim for this special issue on Ap plications in LA and
EDM covers all aspects of data analytic s in supporting
teaching, learning and administ ration for researchers in P-16
education, and the development of t echnology-enriched
formats of instructional deliver y, such as various categories of
blended and online learning. Tradi tionally, the main data
sources of LA and EDM research re ply on the database in the
learning management system (LMS ). The developments of
Internet of Things (IoT) or sens ors, at some levels, make up
the gap of activity tracking outside the LMS. Th e special issue
endeavors to publish research and practice which explores the
applications of LA and EDM by incl uding data sources
outside the LMS, such as open dat a, in classroom devices,
IoT, mobile devices, academic dat a warehouse and other
devices which can track, diagnose an d store learning activities.
This special issue is also inter ested in innovative approaches
of feature extraction, pattern id entification/recognition, dat a
anonymization, modeling and in tervention to support
innovative applications of mac hine learning and deep learning
in education.
All presentations in this special edit ion were referred
through a double-blinded procedu re before being accepted
for publication. Each manuscrip t submitted was reviewed by
two to three invited reviewers. The r eview criteria were:
importance of the subject;
originality of the approach;
soundness of scholarship displayed;
level of interest and pertinence for readers;
depth and strength of argument; and
clarity of expression.
The first article, Youngjin Lee’s“Est imating student ability
and problem difficulty using item res ponse theory (IRT) and
TrueSkill”, examined item respons e theory (IRT) and
TrueSkill applied to simulated and real problem-sol ving data
to estimate the ability of studen ts solving homework problems
in the massive open online course (M OOC). Based on the
estimated ability, data mining mo dels predicting whether
students can correctly solve homewor k and quiz problems in
the MOOC were developed. This stud y found that the
correlation between students’ab ility estimated from IRT and
TrueSkill was strong. In additio n, IRT- and TrueSkill-based
data mining models showed a compar able predictive power
when the data included a large number of stud ents. While
traditional IRT research has be en focused on the assessment
design and development, this st udy explores the application of
IRT in measuring student’s capabili ty by taking advantage of
MOOC.
The second article, Dimi
cGabrijela,DejanRancˇi
c, Nemanja
Macˇek, Petar Spalevi
candVidaDrąsut _
e’s“Improving the
Prediction Accuracy in Blended Learning Environment using
Synthetic Minority Oversampling Technique”, studied the
prediction accuracy of student’s activity patterns in the blended
learning environment. Classification models were used in the
comparison for different cardinality subsets. The results showed
the opposite with the fact that reducing the number of features
leads to prediction accuracy increasing. The authors argued that
improving the prediction accuracy in the described learning
environment is based on applying synthetic minority oversampling
technique what had affected on results on correlation-based
feature selection method. While highly imbalanced data is a
challenge for predictive modeling, this study proposed a method to
enhance prediction accuracy via synthetic minority oversampling.
The third study, Brian Wright’s“Analys is of supportive
campus environments and first-gen erations-student learning
outcomes”, intends to analyze the r elationship between
supportive campus measures an d student learning outcomes
for first-generation students and n on-first-generation students
to determine if variances are pres ent via EDM. Data were
gathered through cluster samplin g of pre-existing datasets on
undergraduate seniors generat ed by issuances of the National
Survey on Student Engagement. It found that clear pattern
differences are present between first -generation and non-first-
generation students in terms of supportive campus
environment factors contribut ing to learning outcomes. While
server log is the major data source in LA and EDM (Ihantola
et al.,2015;Schwen dimann et al.,2017), this study anal yzed a
longitudinal investigation and id entify useful insights for
higher education administrator s.
The fourth article, Riccardo Pecori, Vincenzo Suraci and
Pietro Ducange’s“Efficient computation of key performance
indicators in a distance learning university”,proposeda
framework to compute efficiently key performance indicators,
summarizing the trends of students’academic careers, by using
EDM. The parallel computation of the indicators through Map
and Reduce nodes is carefully described, together with the
workflow of data, from the educational sources to a NoSQL
databaseandtotheLAengine.Theframeworktestedinan
Italian distance learning institution. It concluded the framework
was able to significantly reduce the amount of time needed to
The current issue and full text archive of this journal is available on
Emerald Insight at: www.emeraldinsight.com/2398-6247.htm
Information Discovery and Delivery
47/2 (2019) 65–66
© Emerald Publishing Limited [ISSN 2398-6247]
[DOI 10.1108/IDD-05-2019-074]
65
To continue reading
Request your trial