Students Expectations on Learning Analytics: Learning Platform

Features Supporting Self-regulated Learning

Tuija Alasalmi

School of Vocational Teacher Education, Haaga-Helia University of Applied Sciences, Pajuniityntie 11, Helsinki, Finland

Keywords: Learning Analytics, Self-regulation, Expectations, Online Learning Platform.

Abstract: The article presents results of a survey and interviews mapping students’ expectations and needs on learning

analytics. The discussion focuses on the functionalities and features which were considered to support self-

regulated learning in Moodle learning environment. The aim of the discussion is to identify how the reported

student needs could be met by utilizing descriptive, prescriptive or predictive learning analytics.

It was discovered that students need and expect certain functions in the digital learning environment to support

self-regulated learning. The survey results indicate that students mainly demand for tools which could help

them in planning and scheduling their studies. Secondly, to be able to monitor and regulate their performance,

they need progression tracking tools as well as timely and constructive feedback. Features of descriptive

analytics were considered the most useful for self-regulation, while the expected benefits of prescriptive and

predictive analytics were more controversial and tentative.

1 INTRODUCTION

Learning analytics is a relatively young field of

research, and its use in the field of education is still in

its infancy. Learning analytics is a sub-genre of data

analytics and does not have a definition set in stone.

Learning analytics is broadly defined as the collection,

measurement and analysis of data produced by a

learner in a digital learning environment. The data

processing aims at understanding and optimizing

learning and its contexts (Society for Learning

Analytics Research, 2020).

Although the data in digital learning platforms is

provided by the students, their perspective has mainly

been bypassed in the development of learning

analytics tools (Hooli, 2020; Buckinham, Shum,

Ferguson, Martinez-Maldonado, 2019). Therefore,

learning analytics tools often only measure and record

actions which present useful information to the

teacher, but not necessarily to the students (Hooli,

2020). Moreover, even if students could benefit from

the data collected in the online learning environment,

they may not have access to view it. In recent years,

most digital learning platform providers have started

to develop learning analytics tools which are

available to students, ie. they are able to themselves

https:/www.haaga-helia.fi

monitor and utilize the data accumulated from their

online studies.

The tools for collecting, analyzing, and interpreting

learning data will not develop unless learning analytics

is researched and advanced in a user-driven way. If the

tools are found to be unnecessary or useless, they will

not be deployed. Silvola, Jylkäs and Muukkonen have

concluded that in order to develop analytics tools that

students want to use, the tools should meet the

information and support needs of students in their daily

lives (Silvola, Jylkäs & Muukkonen, 2020). Therefore,

it is important to map the real need of end-users of

learning environments – both teachers and students –

and to critically assess if the data collected really

benefits them.

This article is based on a survey and interviews

conducted between November 2020 and February

2021 in the MOPPA project (Motivation och

självreglering på inlärningsplatta med hjälp av

inlärningsanalytik). The aim of the survey and

interviews was to map students’ experiences of digital

learning environments, identify everyday support

needs for online learning and find out which

activities, tools and pedagogical elements

contributing to the collection of learning data in

digital learning platforms are considered to increase

students’ self-regulation and motivation.

Alasalmi, T.

Students Expectations on Learning Analytics: Learning Platform Features Supporting Self-regulated Learning.

DOI: 10.5220/0010537101310140

In Proceedings of the 13th International Conference on Computer Supported Education (CSEDU 2021) - Volume 2, pages 131-140

ISBN: 978-989-758-502-9

131

On the basis of the survey some key needs were

found in the context of Moodle learning environment

used in Haaga-Helia University of Applied Sciences.

This article specifically discusses the tools, activities

and functionalities which the respondents of the

survey found useful and supportive of their learning

process. The reported solutions (either pedagogical or

technical) which were believed to facilitate learning,

are viewed as either existing or possible data

collection points for descriptive, prescriptive or

predictive learning analytics. Practical suggestions

concerning the development of learning analytics

tools are made on the basis of the findings.

2 DATA ANALYTICS IN DIGITAL

LEARNING ENIVRONMENTS

Learning analytics is commonly divided into three

types: descriptive or diagnostic, prescriptive and

predictive analysis. Gröhn and Nevalainen have made

a comparative analysis of various online learning

platforms with respect to the in-built learning

analytics tools available. They report that the

analytics tools in the platforms included in their

analysis mainly offer descriptive or diagnostic

learning analytics. There are yet only a few tools or

options for using predictive analytics available, and

there’s a lot of variation in the use of prescriptive

analytics (Gröhn & Nevalainen, 2018). Even though

tools and models for predictive analytics such as

identifying likely drop-outs have been developed,

they have not been widely adopted in higher

education (Viberg et el., 2018). Gröhn and

Nevalainen conlude that learning analytics in the

platforms scrutinized in their report shortly means

monitoring the various learning actions, collecting the

course resources and assignments into a

comprehensive view and providing reminders or

notifications on the deadlines and materials to be

studied. To summarize, the data offered to learners

includes their progression on a general level, on a

specific course and with the chosen assignments.

Most analytics tools also offer separate views for

submitted assignments and those yet to be done

(Gröhn & Nevalainen, 2018).

2.1 Previous Studies of Students’

Expectations on Learning Analytics

Schumacher and Ifenthaler have mapped students’

expectations on learning analytics within the

framework of the three phases of self-regulation as

outlined by Barry Zimmerman (2002): the

forethought phase, the performance phase and the

self-reflection phase. In this framework self-

regulation is understood as a set of metacognitive and

cognitive processes initiated and maintained by the

students, to carry out the tasks given in any context of

learning (Zimmerman, 2002). The students’

expectations on learning analytics in the study of

Schumacher and Ifenthaler were recognized as to-do

lists, reminders of deadlines, tools for planning, clear

learning goals, motivational aspects and individual

recommendations (forethought phase); monitoring

competence level and development of skills,

additional or personalized material and assignments,

recognition of offline and social learning

(performance phase); and self-reflective assignments

and feedback (self-reflective phase) (Schumacher &

Ifenthaler, 2018).

Students' expectations towards learning analytics

have been studied in Finland previously in the APOA

project. In the workshops organized for students, the

following ten themes emerged on which learning

analytics is expected to offer support: 1) Defining and

monitoring learning objectives, 2) Monitoring

learning activities, 3) Monitoring and visualizing the

learning process, 4) Interaction, 5) Feedback, 6)

Competence development assessment, 7) Provision

of study material, 8) Influencing the process or the

content of the course, 9) Improving teachers'

technical competence and 10) Time management and

study skills (Hartikainen & Teräs, 2020).

2.2 Moodle

In the Moodle learning environment used in Haaga-

Helia, the analytics tools available to the student are

mostly descriptive. These descriptive tools include

Completion Progress block, which collects and

visualizes information about the completion of

scheduled tasks as well as activities and resources in

any course area. The function is not automatic; the

teacher activates it if desired and makes it available to

students. Tools for collecting and organizing learning

data also include the Gradebook and especially the

personal Dashboard which is a kind of landing page

and summary view that the student can customize and

from which they can see the general situation of all

their courses (including completion percentages of

each course, if progress completion is enabled). The

dashboard includes a link to the site-level calendar

tool which collects deadlines from all the active

courses available to the student. Students can add

various blocks into their dashboard, such as timeline,

upcoming events, progress tracker, and latest

CSEDU 2021 - 13th International Conference on Computer Supported Education

132

announcements. The blocks collect data from all the

course areas in which the student participates.

The range of tools available to the teacher is wider

and there are e.g. log data, reports, and activity-specific

usage statistics that students do not have access to. All

Moodle activities (e.g. forums, test, assignment,

lesson, etc.) collect information about the student's

activity, such as the amount of views. This information

is only visible to the teacher, while the student can only

access the material he or she produces (i.e., assignment

submissions, test attempts, own files, forum posts, and

comments). This information, however, is scattered

and not presented as a single comprehensive course-

level view or dashboard to student.

Possibilities for prescriptive analytics in Haaga-

Helia’s Moodle are limited. The test activity enables

automated feedback and giving instructions to the

student based on the question or test results. There are

also automated reminders used for activating students

to submit their assignments before deadlines.

Teachers may use reports to sort out inactive students

on particular activities and send messages with

further instructions, but all this has to be done

manually. Finally, neither teachers nor students have

any tools available for predictive analytics. The

Moodle Learning Analytics API is installed in Haaga-

Helia’s Moodle, but all the default learning analytics

models are disabled.

3 METHODOLOGY

As one of the purposes of the project was to identify

and explore the connection between student self-

regulation and learning analytics, the framework for

the survey was based on Zimmerman’s model of self-

regulated learning. It was also used by Schumacher

and Ifenthaler in their study of students’ expectations

on learning analytics as reported earlier.

3.1 The Survey

The survey consisted of 25 questions divided into three

question sets. The first seven questions regarded

background info, and the next seven questions mapped

the students’ preferences concerning online learning.

The final set of 14 questions included structured

questions on the students’ perception of their self-

efficacy level and their experiences on certain

pedagogical solutions and functionalities available in

their online courses. There were also open-ended

questions about the perceived barriers to online

learning, the expectations and needs for tools or actions

supporting learning, and two questions on motivational

issues. This article focuses on the responses given in

the open-ended questions; however, the issue of

motivation was omitted from this discussion.

The open-ended questions of the survey were

analysed through qualitative content analysis. The

needs (functionalities or tools) reported by the

respondents were first categorized according to their

main theme, identifying repeating key words and

phrases. Thematic analysis also utilized semantic

analysis. Furthermore, the identified main themes

were recognised as needs towards either descriptive,

prescriptive or predictive analytics.

The survey was sent to selected student groups in

blended learning study programs, but participation

was voluntary and the answers were handled

anonymously. There were 47 responses and most of

the respondents were students of business and

entrepreneurship or IT and digital services. The mean

age of the respondents was 34; the youngest was 21

and the oldest 56. They were mostly adult students

with a previous degree or work experience.

3.2 The Interviews

The survey was followed by voluntary semi-

structured interviews with seven students who had

also participated in the survey. The interviews

included four open-ended question and two semi-

structured question based on the survey results. With

the semi-structured questions, the interviewees were

shown a table that summarized the tools or actions

supporting learning as identified in the survey

responses, divided into the three analytics types. The

summary was made on the basis of the survey

responses, and it included the most often suggested

features and functionalities. However, some options

were added simply because it was presumed that most

students were not fully aware of the possibilities of

learning analytics in learning platforms. Table 1

shows the categories including the titles they were

presented with. The first column represented needs

for descriptive analytics, the second column

represented needs for prescriptive analytics and the

third column consisted of needs for predictive

analytics.

The students were asked to choose which of the

categories included features that best support self-

regulated learning and motivation in general. They

were also asked to highlight any three features which

they personally considered the most useful for them.

The interviews were conducted by the author and

another researcher and they were transcribed. All

seven interviewees chose to have a Zoom or Teams

interview. Oral informed consent to record the

Students Expectations on Learning Analytics: Learning Platform Features Supporting Self-regulated Learning

133

Table 1: The division of actions and features supporting self-regulated learning on the learning platform, categorized under

the types of learning analytics.

session was gained in the beginning of the interview

by all interviewees. The interviews lasted between 15

and 28 minutes. The questions were aimed at

revealing in more detail the reported needs,

experiences and perceptions of the students within the

context of learning analytics. The purpose of the

interviews was not to gain any new information but

rather to find further explanations for the identified

needs and to elaborate on the specific themes that

emerged in the survey. The interviews complemented

the thematic analysis of the survey responses.

4 RESULTS

In an open-ended question of the survey, students

were asked to write about anything they feel could

help them (self-regulate) in their study process in

online learning. Respondents were especially

encouraged to think if they had any needs for a

specific feature or functionality in the learning

platform. Eleven students did not have any

suggestions nor reported specific needs; in addition,

there were some replies focusing on pedagogical

issues. However, two major themes emerged: a

demand for time management and progression

tracking, and the need for feedback.

4.1 Time Management, Progression

Tracking and Study Planning

In 22 responses, a need for scheduling and planning

tools was highlighted, with a strong emphasis on the

follow-up of deadlines:

“A to-do list in one place would be great. I could

add notes myself about submitted assignments and

schedule all undone assignments.”

Another feature or functionality considered

important was a comprehensive dashboard view for

monitoring all the active courses at one glance:

“A comprehensive view of all active courses,

currently I’m building such a view with excel myself.

The view includes all contact sessions (online/

classroom), personal events, deadlines and my own

scheduling of which course assignments I am

working on and when.”

“[A tool] that shows me instantaneously which

courses are ongoing and what happens in them and

how I have progressed and what are for instance the

assignment deadlines, and all of these in one

location.”

The majority of student needs were connected

with progression monitoring. This was needed both

on a single course level and within the framework of

the entire study program:

“It would be nice to see information of my actions

and time usage, and a comprehensive view of my

CSEDU 2021 - 13th International Conference on Computer Supported Education

134

progression in a particular course and within my

studies as a whole. Now this information is scattered

in different locations and in a majority of the courses,

the progression tracking does not work correctly.”

Some students specifically emphasized the

importance of visualizations such as heat maps, visual

timelines and progress bars. One of the respondents

even claimed that learning platforms should be more

like project management systems where students

would be able to manage all learning tasks according

to their own schedule. This sort of Kanban-style view

on learning emerged in the responses quite often and

it was evident that tools for time management are

certainly needed. However, it is interesting that the

responses had a task-oriented edge as they did not

reflect the need for monitoring competence

development, as was found in the student workshops

of the APOA project and in the study by Schumacher

and Ifenthaler. According to Zimmerman, setting

personal goals in the forethought phase of self-

regulated learning, and monitoring one’s

performance in respect to the expected learning

outcomes in the monitoring phase are functions

supporting self-regulation (Zimmerman, 2002).

Nevertheless, this was not highlighted in the students’

needs in this survey. Since the need for keeping track

of deadlines and assignment submissions was

emphasized, it appears that instead of having

problems with achieving the expected learning goals,

the metacognitive difficulties were a bigger issue.

Apparently, some students struggle finding suitable

strategies for online learning and they may have some

problems with self-regulation, but only in terms of

time management. This is where they expect help

form learning analytics.

4.2 Different Types of Feedback

The next important viewpoint after time management

was the question of feedback. 15 students reported

issues especially with communication and feedback,

and it was evident that tools were needed for a more

effective and timely communication concerning the

students’ overall status, course and task level

progression and the level of engagement:

“My work progresses/does not progress would be

a good button to have in all courses, this is how

teachers would know more about the student and they

could give instructions and push students forward in

doing the assignments.”

Such a feature might help teachers in identifying

students who are at greater risk of dropping out or

who may have trouble with their cognitive or

metacognitive processes (i.e. self-regulation). This

sort of need for a process-oriented feedback was,

however, in the minority. The responses emphasized

more of outcome-oriented feedback, which is what

most learning analytics dashboards usually support

(Sedrakyan, Malmberg, Verbert, Järvelä &

Kirschner, 2018). Sedrakyan and colleagues point out

that providing relevant type of feedback in learning

platforms depends on identifying the factors behind

students’ (low) performance; any learning analytics

dashboard should be able to differentiate whether the

student has cognitive problems (difficulties in

understanding the tasks) or behavioural problems

(difficulties in motivation or putting enough effort

into studying) and offer feedback for the purposes of

cognitive development or behavioural changes

(Sedrakyan et al., 2018).

In the survey, most of the respondents hoped for

instant feedback or at least feedback given relatively

promptly after the assignment submission:

“Feedback is needed in every course soon after an

assignment is submitted in order to be able to learn.

In this course feedback was delivered quickly, but this

not always the case.”

“In math course there were assignments which

had instant feedback on whether or not the answer

was correct. I wish there were more of this sort of

gamified assignments.”

Well-timed feedback does not only help students

to self-regulate their learning efforts, but it also helps

students in identifying their competence level and

ultimately it leads to better learning results. For

example, in a study with 500 students carried out by

Liu and Cavanaugh, the teachers’ constructive and

timely feedback comments had significant effect on

the students’ final scores in a math course (Liu &

Cavanaugh, 2012).

It is worth considering a learning analytics

functionality which could sort out students who

would benefit the most from feedback without any

delay, since many of the responses expressed

dissatisfaction with the timing of the feedback.

“The weekly assignments should be checked and

feedback should be received earlier than the end of

the course. Often course material is based on the

previous weeks’ assignments and if I have understood

something wrong or have no competence in the

beginning of the course, there is no way to fix it in the

end.”

Sedrakyan and colleagues have suggested, based

on their research on learning analytics dashboards as

a source of feedback, that the optimal time for

feedback can be determined by detecting any

anomalies or challenges in the learning process

(Sedrakyan et al, 2018). One survey participant

Students Expectations on Learning Analytics: Learning Platform Features Supporting Self-regulated Learning

135

suggested that in connection with the assignment

submissions, there could be an optional button for

asking feedback (especially in tasks were feedback is

usually not given). This could benefit students who

need more guidance while teachers are able to reduce

their work load by not giving feedback to everyone

on all tasks. It has been discussed that as students

differ in their readiness for self-regulatory learning,

support should be offered to the students with weaker

metacognitive skills (Nussbaumer et al., 2015).

Regulating the level and receive time of feedback

according to the learner’s needs is what learning

analytics dashboards could be used for, as Sedrakyan

and colleagues have also proposed earlier (Sedrakyan

et al, 2018).

4.3 Behavioural or Cognitive

Feedback?

Earlier studies have outlined the types of cognitive

feedback into corrective, epistemic and suggestive

feedback (Alvarez, Espasa, Guasch, 2012; Guasch,

Espasa, Alvarez & Kirschner, 2013). Corrective

feedback points out false conclusions, solutions and

decisions; epistemic feedback provides analysis for

critical reflection and suggestive feedback gives

advice to the learner on how to proceed and what to

improve. All feedback types may play a role in

supporting self-regulated learning, but from the point

of view of learning analytics, corrective and

epistemic feedback types are more descriptive and

suggestive feedback contributes to prescriptive

analytics. Automated feedback in learning platforms

can form a basis for prescriptive analytics for the

students – at least if the feedback is accompanied with

suggestions or instructions on what to do next on the

basis of the results.

Many of the survey responses regarded feedback

as more of a verification of the task being completed

successfully, and it can be concluded that feedback is

an integral feature of descriptive learning analytics.

Receiving correct answers is considered important, as

is the descriptive assessment of what went well and

what went wrong, as one student responded. The need

for instant, corrective feedback was evident, but it

remained a bit unclear whether this type of feedback

was considered cognitive or rather more behavioural.

Therefore, feedback was chosen as one of the key

topics in the interviews and the interviewees were

asked to describe their notion of good feedback in

more detail. They were also asked to define if the

source of feedback (teacher, peers or learning

platform (automated or auto-generated feedback))

had any significance to them.

The responses were quite univocal: good feedback

is well timed, constructive (cognitive) and supportive.

The first two characteristics were equally important but

the difference between behavioural and cognitive

feedback clearly emerged. Behavioural feedback

should be instantaneous, as one interviewee described:

“In fact it is not really feedback, it is connected to

what came up previously, that when I registrate on a

course, I don’t actually know if I was accepted or not.

[…] That gives no confirmation or such. […] For

myself, I prefer receiving the feedback in this

situation directly from the system. Like “great job,

this is now completed, it is done”. (Interviewee 1,

female)

There were three other interviewees who brought

up similar viewpoints describing instant, automated

feedback:

“It can be continuous feedback, like whenever

you submit an assignment, you would get “Great, xx

% of the course is now completed”. It may be generic,

but it still leaves a good feeling.” (Interviewee 5,

female)

Many of the interviewees distinguished the

automated feedback in the platform as being more

behavioural and the (delayed) manual feedback given

by the teacher as more cognitive and important:

“The correctiveness is probably the priority, in a

sense that if it comes afterwards, at least you know

for the future what went wrong even if the feedback

did not help you in the learning process during the

course”. (Interviewee 4, female)

The responses imply, as has been concluded in

earlier research, that behavioural feedback is just as

important as cognitive feedback because it provides

students opportunities to reflect on the learning

process, thus helping students self-regulate their

learning (namely, monitoring, planning and adapting)

(Sedrakyan et al., 2018).

Cognitive feedback is a bit more flexible in terms

of perfect timing, but nevertheless the time between

assignment submission and receiving feedback

should not be too long as it has a negative effect on

self-regulation, especially if actions are expected on

the basis of the feedback. An interviewee described

the mismatch between personal schedules, deadlines

and received feedback, which led her to drop out of

the course:

“I got constructive feedback of which I realized that

I have to a lot to correct, but right now I am really busy

at work and I have no time to make the corrections.

[…] It was like, the (final) deadline in this assignment

had to be set by myself and I had set it for the day after

tomorrow. I got the feedback a week earlier, so I just

don’t have the time [to make the corrections]. If there

CSEDU 2021 - 13th International Conference on Computer Supported Education

136

had been a possibility to postpone the deadline, then of

course.” (Interviewee 7, female).

The perfect timing of the feedback thus seems to

depend on the quality and content of the feedback.

Another student suggested that the date when

feedback is available should be announced in the

learning platform, to be able to plan the study process

in the course to match with the schedules of personal

life. Such a feature is probably quite easy to add in the

settings of the activity tools in online platforms, and

it could contribute to the data being collected into

learning dashboards for progress monitoring.

The type and purpose of the feedback also affects

the preferences on the feedback source. Cognitive

feedback is expected from the teacher on specific

assignments and overall competence development.

Some suspected that artificial intelligence could not

offer as good feedback as a teacher. The reason was

that the teacher was considered an expert, and the

view of an expert was highly appreciated. Similarly,

if feedback is merely behavioural and confirmative by

nature, it can be automated or auto-generated by the

learning platform:

“I don’t think there is a problem with whether the

feedback comes from the machine or the teacher. It

depends on the quality of the feedback, of course. It

is not certain that you’ll get more than “ok” from

other students or the teacher. So if it [feedback]

comes from the computer or the teacher, it doesn’t

matter.” (Interviewee 6, male)

“It depends on the type of feedback. If you only

do a test in which you choose alternatives, it is ok to

get the feedback from the system. No further

comments are needed. But if the assignment is a

written essay with reference material, it is nice to get

written feedback, not just a numeral grade.”

(Interviewee 3, female)

Nearly all interviewees claimed that it is good to

have multiple sources of feedback. Interestingly

enough, supportiveness was expected mostly from

peer feedback. Peer feedback was also considered as

part of the learning process or as a task for practicing

cognitive skills such as argumentation:

“A few courses have had peer review practices

and with fellow students, they don’t necessarily have

the skills to give feedback. It should be trained more

before practicing it.” (Interviewee 6, male)

Therefore, constructive feedback was not really

expected from peers. The benefits of getting positive

confirmations and different viewpoints were, however,

acknowledged. Peer feedback was suggested to be

incorporated in the learning platform with simple and

easy ways, such as giving star ratings or thumb-ups.

Students preferred fast and effortless ways of giving

peer feedback, since group and individual assignments

were considered arduous and time-consuming.

The importance of peer evaluation lies within co-

regulated learning as students learn to regulate their

efforts when they are being compared to others’

behaviour and outcomes (Sedrakyan et al. 2018).

However, in the survey, only 15% of the respondents

thought it would be useful to have a comparative

report of their progression and competence

development with respect to their peers. 60 % replied

that such a view would not benefit them and 25 %

were not sure if such a function was useful. The issue

was discussed by some of the interviewees; one

student felt peer comparison dashboards might work

because of the competitive element, but another

student regarded it as possibly disturbing from the

point of view of self-regulation.

4.4 Descriptive Analytics: A Must Have

When shown the table 1 during interviews, all of the

students selected descriptive analytics as the most

important and useful category from them. One student

even claimed that descriptive analytics is a norm and

should be available by default, so it should not be a

matter of choice or preference. Another student chose

descriptive analytics as the most important in the

context of independent online studies, but claimed that

in blended learning course, prescriptive analytics is

needed more. However, this overall opinion does not

fully match with the features which the students

selected as their favourites. The summary of their

choices is described in table 2; the number of students

who selected the item is shown in brackets.

There were two choices made outside the

presented table: the other one was an access point for

all material on a single course and the other was a

view based on received feedback (how well the

student had succeeded as compared to the assignment

requirements). These two choices may, however, be

categorized into features of descriptive analytics.

To be able to view and monitor what is done or

undone was the most important feature, as the survey

responses also revealed. Currently, in Haaga-Helia’s

Moodle such as view is offered on a course level (the

course gradebook and on some courses the

progression tracking block). Within the context of the

whole study program, monitoring of one’s

progression is not available in Moodle and that was

one of the issues criticized by the students. They feel

it is not easy to log into another system – or at worst

into multiple systems – to get a comprehensive view

of their studies. This sort of overall view could be

constructed with Moodle’s Learning Plan tool,

Students Expectations on Learning Analytics: Learning Platform Features Supporting Self-regulated Learning

137

Table 2: The division of actions and features supporting self-regulated learning on the learning platform, the amounts of

student selections in brackets.

complemented by the Competency tool. Or

alternatively, a plugin or an integrated external

software could be used for offering a comprehensive

student dashboard for study progress monitoring on a

study program level. On a course level, the current

tools for tracking progression seemed to be enough,

but students hoped the tools would be used regularly

and consistently in all study units.

Other preferred features of the listed descriptive

analytics were the assignment notifications and view

of upcoming (and past) deadlines. The responses do

not clearly indicate if the Moodle calendar in student

dashboard is in efficient use; but at least it is possible

to use the calendar as a tool for planning and

gathering all the deadlines from all courses in one

location. Students may also mark their own events in

the calendar.

Descriptive analytics and the need for it was

widely discussed and explained by the students in the

interviews. For example, the possibility of monitoring

one’s time usage was an interesting option for most;

clearly it was considered as a potential way of

regulating the personal study process and efforts:

“I like to analyse how I’ve spent my time.

Information on what I’ve used my time for in the

course and how long I’ve spent moving around in the

environment could help me in defining if I have used

my time reasonably.” (Interviewee 7, female)

“Sure, it would be interesting to see how long I’ve

used for a particular task. In my everyday life there is

quite a many interruption.” (Interviewee 6, male)

Another student was more suspicious of the

ability of learning analytics to provide accurate study

time data, simply because she regularly used to

download the course material and read it offline.

Furthermore, being able to identify challenging tasks

or materials was considered useful in principle, but

the students suspected that in practice it would be

difficult to define reliable indicators. As one of the

students remarked, used time and the amounts of

clicks do not necessarily correlate with challenges or

difficulties in the learning process.

4.5 From Descriptive to Prescriptive

Analytics

The interviewees selected suggestions for suitable

study paths and methods of progression as the second

important functionality. Evidently, there is a need for

prescriptive analytics, but this may be due to two

reasons. First, in the survey responses, the courses in

Haaga-Helia’s Moodle were generally considered to

have a bad structure and messy layout. If the course

view is experienced as incoherent, it is

understandable that there is a need for a tool that

could pick out and organize all the important study

material and assignments:

CSEDU 2021 - 13th International Conference on Computer Supported Education

138

“[…] everything that is connected to the course

would be on different pages and then it would

progress page by page or whatever the logic would

be. But it would be sequenced in a way that when you

can do this, you may move forward to this [material].”

Secondly, many students were very busy with

their work and family, and they expected the learning

platform to somehow help them with scheduling by

offering and suggesting ways to proceed with course

selection, for instance. Therefore, the need for

prescriptive analytics is connected with time

management and self-regulation processes, as one of

the interviewees described:

“I have quite a demanding job, I’m in a hurry at

times. To be honest, it is sometimes challenging

because I don’t have the time to focus on which

courses I have to work on and such. So, if there’s

some sort of guidance, as I said in the beginning, or

recommendation available.” (Interviewee 1, female)

Even though the features of prescriptive analytics

were not generally seen as important as the

descriptive features, the tools for guidance and

steering the study process were discussed extensively

by many of the interviewees. For example, when

asked about using chat bots in student advising and

getting further instructions, one of the students

replied:

“It’s essential that I get to go forward. Another

thing as a blended learning student is that I do

assignments and advance my studies quite late in the

evenings or weekends. I don’t presume teachers work

at that time. If a machine can help me, then there’s

additional value to me.” (Interviewee 4, female)

Some students were more prejudiced towards the

usefulness of prescriptive analytics based on

automation:

“I was thinking about getting tips or instructions

based on my learning results. In principle that sounds

fine. But when I think of what Spotify suggests to me

based on what I have listened, it doesn’t really match.

I am a bit sceptical.” (Interviewee 7, female)

Nevertheless, prescriptive analytics is what

should be developed alongside descriptive analytics,

since mere description of performance does not

enhance learning. Tanes and colleagues (2011)

analysed the content of feedback messages which

were sent to students at risk of dropping out, on the

basis of the data signals recorded in the system. They

found out that the messages contained summative

analysis without any instructions or feedback on the

learning process or suggestions on how to overcome

the challenges. Consequently, the messages had no

effect on students’ learning (Tanes et al, 2011).

4.6 Accurate Predictions for

Optimizing the Study Process

Predictive analytics was met with a dual response:

one interviewee considered it the most promising type

of learning analytics where we should head for and

another student thought it mainly helps teachers.

Students also had mixed opinions on predictions or

estimates on study time: some believed it cannot be

accurately predicted because learning strategies,

styles and conditions of studying vary; and others

thought it could help a lot with self-regulation simply

because they would be able to plan how much time

they should reserve for studying. An estimate on

expected reading time or time required for completing

an assignment, a task or the whole course was seen as

a way to make time usage more effective, since it

could be easier to optimize regular studying and

balance it with personal life by using free time slots

efficiently. One of the interviewees suggested a

predictive thermometer tool which could constantly

monitor the study progression on a course level,

providing alerts when the study pace is too slow,

giving suggestions on what to do and demanding

actions and rescheduling if deadlines seem to

approach too fast. Another student speculated that

grade predictions could actually help students set

reasonable learning goals:

“A recommendation feature could be… when a

task has not progressed… if I could set the expectation

for myself, that I go for grade 4 or 5 in this course and

the system predicts that grade 3 is more likely with this

effort. It could recommend additional material for

reading.” (Interviewee 6, male)

Providing predictions on the risk of dropping out

were viewed as potentially beneficial, but only if a

possibility for extra guidance and academic advising

was offered in connection with such predictions. It is

possible that some students would take predictions in

a negative way, as one of the students reflected.

5 CONCLUSIONS

Students seemed to appreciate and expect a wide

range of descriptive analytics available to them. In the

survey responses, the demand for time management

and planning tools was significant. Getting relevant

feedback on time was another major need. On the

basis of the interviews, behavioural feedback

contributing to descriptive learning analytics should

be instantaneous and confirmative and this can be

given by the learning platform or the teacher.

Cognitive feedback was expected to be well timed,

Students Expectations on Learning Analytics: Learning Platform Features Supporting Self-regulated Learning

139

corrective and to some extent also suggestive, and it

was expected mostly from the teacher. Supportive

feedback was expected from peers. When the results

are viewed in the context of the three phases of self-

regulation, it is notable that learning analytics tools

are expected to provide help mostly in the forethought

phase (especially with planning and scheduling). The

support from learning analytics in the performance

phase through comprehensive monitoring and instant

feedback is also appreciated and expected. The need

for learning analytics tools in the self-reflection phase

was less evident.

Of the three types of learning analytics

dimensions, descriptive analytics was considered the

most important and even fundamental. Features of

prescriptive and predictive analytics were met with a

more dubious attitude. The scepticism may be due to

the fact that there are only a few prescriptive and

predictive uses of learning analytics available.

However, as Park and Jo remark, as descriptive

analytics begins to be widely available, it is only

natural to add some cases of predictive analytics into

learning platforms and the student dashboard views

(Park & Jo, 2015). Perhaps the prescriptive analytics

could begin with simple recommendations and subtle

suggestions with comparisons such as “students who

read this material, also watched these videos…” or

“students who got the best grades spent 10 hours

reading this material”. In any case, behaviour-based

student dashboards provide important information to

the students alongside knowledge-based dashboards

(Auvinen et al. 2015), and sequential or procedural

analysis of the student’s actions in the learning

process provide data that could help students find

suitable strategies for self-regulation (Sedrakyan et al,

2018). Learning analytics tools should utilize a

mixture of behavioural and knowledge-based data in

order to provide meaningful descriptive dashboards,

useful and well-timed prescriptive analytics and

feedback as well as reliable predictions on learning.

REFERENCES

Auvinen, T., Hakulinen, L. & Malmi, L. (2015). Increasing

students’ awareness of their behavior in online learning

environments with visualizations and achievement

badges. IEEE Transactions on Learning Technologies,

8(3), 261-273.

Alvarez, I., Espasa, A. & Guasch, T. (2012). The value of

feedback in improving collaborative writing

assignments in an online learning platform. Studies in

Higher Education, 37, 378-400.

Buckinham Shum, S., Ferguson, R. & Martinez-

Maldonado, R. (2019). Human-Centred Learning

Analytics. Journal of Learning Analytics, 6(2), 1-9.

Gröhn, A. & Nevalainen S. 2018. Oppimis-

analytiikkajärjestelmät. Ominaisuuskartoitus.

Publication of the eAMK project.

Guasch, T., Espasa, A., Alvarez, I. & Kirschner, P. A.

(2013). Effects of feedback on collaborative writing in

an online distance education. Distance Education, 34

(3), 324-338.

Hartikainen, S. & Teräs, M. (2020). Käyttäjätarpeiden

kartoitus. Opiskelija-opettajatyöpajat. In Hartikainen,

S., Koskinen. M., & Aksovaara, S. (eds.). Kohti

oppimista tukevaa oppimisanalytiikkaa

ammattikorkea-kouluissa. Jyväskylä University of

Applied Sciences, Publication 274. Jyväskylä

University of Applied Sciences.

Hooli, H. (2020). Students’ experiences of learning

analytics in academic advising for supporting self-

regulated learning. A Master’s Thesis. Oulu:

University of Oulu.

Liu, F., & Cavanaugh, C. (2012). Factors influencing

student academic performance in online high school

algebra. Open Learning: The Journal of Open, Distance

and e-Learning, 27(2), 149–167.

Nussbaumer, A., Hellman, E-C., Gütl, C. & Albert, D.

(2015). A competence-based Service for Supporting

Self-Regulated Learning in Virtual Environments.

Journal of Learning Analytics, 2,1, 101-133.

Park, Y. & Jo, H. (2015). Development of the Learning

Analytics Dashboard to Support Students’ Learning

Performance. Journal of Universal Computer Science

21, 1, 110-133.

Schumacher, C & Ifenthaler, D. (2018). Features students

really expect from learning analytics. Computers in

Human Behavior, 78, 397-407.

Sedrakyan, G., Malmberg, J., Verbert, S., Järvelä, S. &

Kirschner, P. A. (2018). Linking Learning Behavior

Analytics and Learning Science Concepts: Designing a

Learning Analytics Dashboard for Feedback to Support

Learning Regulation. Computers in Human Behavior,

107. https://doi.org/10.1016/j.chb.2018.05.004

Silvola, A., Jylkäs, T. & Muukkonen, H. (2020, August 20).

Opintopolku palvelupolkuna AnalytiikkaÄly-

hankkeessa. Pedaforum presentation. https://analytiik

kaaly.fi/wp-content/uploads/2020/09/SilvolaJylkas-

Muukkonen_PF2020_Opintopolku_palvelupolkuna_A

A_hankkeessa.pdf

Society for Learning Analytics Research (SoLAR). (2020,

November 11). What is Learning Analytics?

https://www.solaresearch.org/about/what-is-learning-

analytics/.

Tanes, Z., Arnold, K. E., King, A. S., & Remnet, M. A.

(2011). Using signals for appropriate feedback:

Perceptions and practices. Computers & Education,

57(4), 2414–2422.

Viberg, O., Hatakka, M., Bälter, O., & Mavroudi, A.

(2018). The current landscape of learning analytics in

higher education. Computers in Human Behavior, 89,

98–110.

Zimmerman, B. (2002). Becoming a Self-Regulated

Learner: An Overview. Theory Into Practice, 2, 63-144.

CSEDU 2021 - 13th International Conference on Computer Supported Education

140