In recent years, university courses have gone through many changes in terms of teaching and assessment methods. The introduction of continuous assessment is one of such changes, encouraging students to carefully plan and spread their e orts over the whole duration of the course. However, this assessment method can imply problems such as task overload, which complicate the elaboration of e ective work plans by students. Ultimately, a poor work plan will likely lead to underwhelming performance by the student. This paper describes a work in progress on how learning analytics can be used in order to help students improve their performance in a continuous assessment setting. Some outlines for the objectives to be ful lled by future work are provided as well.
Over the last decade, the educational eld has been heavily in uenced by the
evolution and progressive popularization of learning technologies. This led to
the birth of several di erent disciplines with the objective of supporting and
enhancing the learning process. Learning analytics (LA) is one of such
disciplines, being the result of applying data analytics techniques to the educational
environment [
This document proposes the use of learning analytics techniques in order to improve students' performance in a continuous assessment setting. The paper starts with a brief overview on learning analytics and dashboards, as well as modern teaching and assessment methods which are used in higher education institutions. Afterwards, an application of learning analytics is proposed with the objective of evaluating and mitigating the interferences of concurrent courses under continuous assessment. Finally, some guidelines are provided to further explore this line of work in the future.
Copyright © 2018 for this paper by its authors. Copying permitted for private and academic purposes 2
Learning analytics is commonly de ned as \the measurement, collection, analysis
and reporting of data about learners and their contexts, for purposes of
understanding and optimizing learning and the environments in which it occurs" [
As e-learning features became more present in educational scenarios, the
solutions provided by learning analytics grew more sophisticated and e ective.
Particularly, the now generalized use of learning management systems (LMS)
vastly facilitates the collection of student data for the purpose of analysis. In
its most basic de nition, a LMS is \a software application that automates the
administration, tracking and reporting of training events" [
The main element responsible for the visualization of analysis results and
indicators is the learning dashboard. This tool is often embedded into the LMS,
although it can also be a standalone utility. The goal of a learning dashboard
is to \capture and visualize traces of learning activities, in order to promote
awareness, re ection and sense-making, and to enable learners to de ne goals
and track progress toward these goals" [
Members of the LA research community have developed their own learning
dashboards as part of their work, implementing some unique functionalities.
These features may be directed to students [
A properly designed dashboard is often necessary to convey the results of data analysis in education to the target audience. Easily comprehensible tables and graphs are normally the most e ective way to display the desired information. 3
As stated earlier, the in uence of new technologies was the source of numerous changes observed in teaching and assessment methods during recent years. This led to events such as the quick rise in popularity of massive open online courses (MOOCs), as well as deep transformations in higher education institutions.
Many university courses nowadays implement some form of blended learning,
an education program that combines traditional in-classroom teaching with the
delivery of online resources and activities to the student [
The introduction of blended learning spawned several di erent teaching
models that combine both brick-and-mortar and online learning. One of the most
popular is the ipped classroom model, which is de ned as \an educational
technique that consists of two parts: interactive group learning activities
inside the classroom, and direct computer-based individual instruction outside the
classroom" [
Alongside teaching methods, there have been some recent developments in
assessment methods as well. Continuous assessment is now very common in many
Spanish universities as encouraged by the Bologna process [
A course under the continuous assessment method normally contains a series
of formative and summative tasks. The former are low-risk activities that are
meant to follow the students' progress with no impact over their nal grade,
while the purpose of the latter is evaluating their knowledge through an activity
or exam that counts towards the nal grade [
4.1
The purpose of this work is providing solutions to improve the e ectiveness of continuous assessment using learning analytics techniques. This is a long-term project of increasing complexity.
The rst speci c objective of this line of work is studying the interferences between courses under continuous assessment that are carried out at the same time. In a university setting, a student is often enrolled in many courses at the same time, which may mean that at some point of the semester they must take on a large number of tasks belonging to di erent subjects in a short period of time. In said circumstance, the student has to adopt a certain strategy: it is their decision whether to split their attention evenly among all subjects, or prioritize some over others. Looking at their nal performance, the magnitude of the interferences can be determined, as well as the students' chosen preferences.
This study will focus on students during their rst semester at university. Since they are not familiar with the continuous assessment used at university, they are more likely to have a harder time planning their e orts due to inexperience, possibly accentuating the e ects of interferences.
The following data is available, corresponding to rst year students from University of Vigo's Telecommunications Engineering School: { Anonymized grades from one subject, including all of the summative tasks performed throughout the course. { The dates on which summative tasks take place, corresponding to all concurrent subjects, along with the weight of each task | that is, how much they contribute towards the nal grade.
Since grade data is anonymized, it is not possible to perform an e ective study targeting individual students. Instead, the analysis will be performed in a group-wise manner: as there are too many enrolled students, they are split into several smaller groups, each one of them having weekly in-classroom sessions at a di erent time and day of the week.
Analysis tasks on the previously mentioned data will aim to provide an answer to the following questions: { Do interferences between subjects under continuous assessment really exists? If so, how important are they? In other words, evaluate the correlation between performance | grades | and the existence of summative tasks from other subjects. { Are there relevant performance di erences between groups in the same course? If so, do this di erences correlate with the day of the week on which inclassroom sessions take place? { Do students prioritize some subjects or tasks over others? If so, does it hold a correlation with the weight of each task? { Does any pattern exist regarding students that drop continuous assessment?
The conclusions obtained from this study will be used as feedback for both instructors and students towards future courses, namely: { Teachers from concurrent subjects could coordinate their schedules better, in such a way that task combinations that are observed to be very di cult to handle for students can be minimized or avoided altogether. { Students will have valuable information that will allow them to improve their e ort planning throughout the duration of the entire course. 4.2
In addition to the previously described study, an interactive online tool will
be developed in order to help instructors track the progress of their students
in a course where continuous assessment is applied. This tool will be designed
as an add-on to the e-assessment platform BeA (Blended e-Assessment) [
The BeA platform contains a wide variety of functionalities with the
purpose of designing and evaluating exams, as well as providing a communication
method between students and instructors for revision requests on said exams.
This platform is in active development: new features that were added over time
include automatic assessment of multiple choice type exams and, as shown in
Fig. 1, graphical representations that facilitate tracking a student's performance
and compare it to that of their peers [
The new add-on will include a calendar-like widget displaying the schedule of tasks and exams of concurrent subjects, in such a way that time periods with a high load of work for students are easily identi able. Additionally, real-time information about students' progress will be provided, focusing on the di erent groups in one subject. This information includes: { Evolution of grades over time. { Comparison between groups.
{ Tracking of dropout rate. 5
The described work in progress attempts to provide a useful application to address a common problem that students face in current Spanish universities: the need to handle many concurrent subjects under continuous assessment. The proposed solutions will be tested in rst year courses belonging to University of Vigo's Telecommunications Engineering School, and the obtained results will be analyzed and documented later on.
As a long-term line of work, there are more objectives to be met in the future. The rst one of them would be the individualization of the features described above. That is, instead of targeting an entire group of students, provide detailed analytics for single students, allowing for more speci c analysis and advice. As this would imply working with non-anonymized data, students need to provide explicit permission in order to use their personal data for this purpose.
If individual analysis is achieved, then it could be possible to add the di culty of tasks as a factor that could in uence students' decisions in a continuous assessment setting. However, unlike the date or weight of a certain task, its di culty is subjective. Even if the instructor can provide an estimation of how di cult a task is, each individual student may nd it harder or easier depending on personal factors such as preference of previous knowledge. Directly obtaining the students' opinion would be a way of circumventing this issue. In any case, it is clear that e ectively using task di culty as input data is more challenging than utilizing the data types previously discussed.
One feature of the e-assessment platform BeA allows teachers to directly assign seats in the classroom for each student. This is shown in Fig. 2, where the seats, arranged by rows and columns, are represented by colored circles. Seats that have already been assigned to a student are represented by a picture of said student. On the other hand, seats that are free and available to be assigned are represented by green circles, and gray circles mean that the seat cannot be assigned due to a problem unrelated to the course, such as it being broken.
The seat assignment system is currently used for in-classroom exams, giving each student a predetermined spot in order to speed up the preparation time. However, this idea could be extended to regular instruction in a ipped classroom environment, where students are divided in groups for problem-solving activities. With a process of analysis of the strengths and weaknesses of each student, they could be strategically arranged for in-classroom sessions. This way, students in groups could cover each others' weaknesses, making it possible for them to more e ectively learn from each other.
Another goal that will be pursued is adding xAPI compatibility in order to improve the interoperability of the developed tools. xAPI is a speci cation for learning technologies that makes it possible to trace learning experiences of any kind in the form of subject-verb-object statements1, storing them in a learning record store (LRS). One of the advantages of using this speci cation is that any compatible tool should be able to read records stored in this format, allowing for easy data exchange between di erent entities. By de ning a xAPI pro le for data recorded in the scope on this project, the information could be fed to other tools, such as an independent LMS.
This research has been funded by the Galician Regional Government under projects ED431B 2017/67 and ED431D 2017/12. Additionally, this work has received nancial support from the Xunta de Galicia (Agrupacion Estratexica Consolidada de Galicia accreditation 2016-2019) and the European Union (European Regional Development Fund - ERDF). Learn2011), 1 https://xapi.com/overview/