UIVLP: An Improved User Interface and Visualization Technique to

Visualize Learners’ Performances

Mukesh Kumar Rohil

and Trishna Paul

Department of Computer Science and Information Systems, Birla Institute of Technology and Science, Pilani, India

Keywords: User Interface, Visualization, Clustering, Principal Component Analysis, Scatter Plot.

Abstract: In most of the educational setups, the grading of students’ performance is based on their relative standing in

the class. In this work, we develop and present a user interface to visualize students’ performance, expressed

in terms of marks, out of same maximum marks for each subject, scored by the students in various evaluation

components for a subject. First, we statistically select three most informative subjects for the whole class and

then find the individual student’s average score in all components along with the overall average of whole

class for an evaluation component. We assume that the three courses’ performance for which the 3D

visualization is required, is either specified by the evaluator or selected by the system basis principal

component analysis. The visualization procedures have been developed for both, the individual student and

the entire class. The interactive 3D visualization and the bar-graphs can be compared side-by-side and we

visually observe that the scatter-plot of clusters provides better insights as compared to the conventional bar-

graphs. We also observe that the proposed visualization is better than the bar-graphs basis no-reference

BRISQUE image quality assessment. However, there may be certain situations when both types of graphs

might be needed.

1 INTRODUCTION

YA data collection can consist of scalar numbers,

vectors, higher-order tensors, or any mix of these data

types. Data sets can exhibit either two-dimensional or

multi-dimensional characteristics. Color coding is a

singular method for representing a collection of data

visually. Other methods encompass contour plots,

graphs, charts, surface renderings, and depiction of

volume interiors. Furthermore, the integration of

image processing techniques with computer graphics

is employed to provide a multitude of data

visualizations (

Hearn & Baker, 2015; Johnson &

Wichern, 2007

The objectives of scientific visualization include:

1) Investigating and utilizing data and information, 2)

Improving comprehension of concepts and processes,

3) Acquiring novel (unanticipated, profound)

insights, 4) Presenting essential characteristics

effectively by rendering the unseen visible. 5)

Ensuring the accuracy and reliability of simulations

and measurements, 6) Enhancing scientific output

https://orcid.org/0000-0002-2597-5096

https://orcid.org/0000-0003-2823-0812

and efficiency, and 7) Facilitating communication

and collaboration among researchers (

Hearn & Baker,

2015; Johnson & Wichern, 2007

In the present work we propose a technique to

visualize students’ performance in a course by

clustering the display around the average (mean) of

the marks obtained by student in various evaluation

components assuming that each evaluation

component has been evaluated out of same maximum

marks. In many universities the relative grading for a

course is done based on the total marks obtained by

the student. For this purpose the evaluator uses

histogram of the marks (i.e. count of students scoring

a particular marks is arranged in increasing or

decreasing order of marks). The histogram is plotted

as bar graph. We suggest to draw the marks of

students in three subjects or topics and three areas are

selected as the top three principal components.

After this short introduction, subsequently the

paper has been structured as follows. The Section 2,

Related Work provides a review of the related

literature, The Section 3, summarizes the Theory

Rohil, M. K. and Paul, T.

UIVLP: An Improved User Interface and Visualization Technique to Visualize Learners’ Performances.

DOI: 10.5220/0013191000003890

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Conference on Agents and Artiﬁcial Intelligence (ICAART 2025) - Volume 3, pages 861-867

ISBN: 978-989-758-737-5; ISSN: 2184-433X

861

related to the purpose of clustering in the context, The

Section 4 describes the problem, methodology,

implementation and experimentation. The Section 5

describes the results and finally the Section 6,

concludes the findings and lists scope for future

research.

2 RELATED WORK

The progress in human-computer interaction has led

to the development of novel approaches for

examining graphical data in a dynamic manner,

allowing users to have adaptable control. Although

the majority of this research focuses on the

presentation of statistical data, there has also been

significant collaboration with advancements in

information visualization as a whole. This is

especially true for the representation of extensive

networks, hierarchies, databases, and text, where the

difficulties of handling massive amounts of data

persistently arise (Hearn & Baker, 2015; Al-Barrak & Al-

Razgan, 2016).

The field of statistical graphics encompasses the

creation of various contemporary methods for

visualizing data, including bar and pie charts,

histograms, line graphs, time-series plots, contour

plots, and other techniques. Thematic cartography

evolved from individual maps to extensive atlases,

which portrayed data on diverse subjects such as

economics, society, ethics, medicine, and physical

features. This advancement also offered innovative

methods of representing information through various

symbols (Hearn & Baker, 2015; Johnson & Wichern,

2007).

Most of the work related to the visualization of

the students’ performance are focusing on the user

interface for the students to visualize their

performance rather than helping the evaluator to

visualize the insights in the dataset of the marks.

These simply displays the marks in 3D or 2D without

performing principal component analysis. Some

works related to visualization of data mining and

predictions of the students’ performance (Al-Barrak &

Al-Razgan, 2016; Misailidis et al., 2018) are helping the

both the students and the evaluators. The work done

by Humphries et al. (2006) helps the students to

visualize their grade as their performance and the

work by Deng et al. (2019) is course specific and does

not combine more number of related or selected

courses.

Most of the learning analytics tools and discussed

in (Darcy, 2022; Paolucci et al., 2024; Mukred et al., 2024;

Atif et al., 2013) displays bar graphs, pie-charts etc.

depicting the distribution of learners’ performance

including performance improvement (or degradation)

over time, but these tools do not display 3D scalar and

vector plots for most discriminating courses of study.

3 PURPOSE OF CLUSTER

ANALYSIS

Cluster analysis aims to condense a vast dataset into

significant subgroups of individuals or things. The

division is achieved by categorizing the objects based

on their similarity across a predetermined set of

parameters. Anomalies pose a challenge to this

methodology, frequently arising from an excessive

number of extraneous factors. It is essential for the

sample to accurately reflect the population, and it is

preferable for the components to be independent of

each other. There are three primary clustering

techniques: hierarchical, which follows a tree-like

procedure suitable for smaller data sets; non-

hierarchical, which necessitates specifying the

number of clusters in advance; and a hybrid approach

that combines both methods. The development of

clusters is guided by four primary principles:

distinctiveness, accessibility, measurability, and

profitability (sufficiently significant to have an

impact).

In the present work for 3D visualization we

cluster the marks about the point in 3D representing

the mean of scores in three subjects. These three

subjects are selected by Principal Component

Analysis (Johnson & Wichern, 2007).

4 PROBLEM, METHODOLOGY,

AND IMPLEMENTATION

The problem dealt in this paper is a multivariate

problem so that students’ performance can be graded

using this. Cluster analysis technique is employed to

solve this problem.

A. Problem Description

The problem is to graphically represent marks

obtained by different students. We take a case of four

students. Each student registers in three different

subjects. Each student attempts fixed number of tests,

given by the instructor, in each of the three subjects.

So input for the problem is four text files, one for each

student namely student1.txt, student2.txt,

student3.txt, student4.txt. In other words all the data

related to marks obtained by a particular student is

ICAART 2025 - 17th International Conference on Agents and Artiﬁcial Intelligence

862

stored in a single text file. Thus there are four

different files for four students. In each text file

numerical data is provided in a matrix form. Row1

represents test1 (i.e. evaluation component 1) marks,

Row2 represents test2 (i.e. evaluation component 2)

marks, and so forth. Similarly, column1 represents

subject1, column2 represents subject2 and so forth

(please refer tables I to IV).

B. Input

Input data for the problem is presented in the form of

tables below. Each text file corresponding to

particular student is presented as a table i.e. Each

table corresponds to a particular student.

Table 1: Marks of Student 1.

Subject 1 Subject 2 Subject 3

Test 1 250 150 70

Test 2 250 250 250

Test 3 270 250 90

Test 4 310 150 50

Test 5 280 170 50

Above table shows the marks of student1 in three

different subjects i.e. subject1, subject2 and subject3.

Table 2: Marks of Student 2.

Subject 1 Subject 2 Subject 3

Test 1 0 100 20

Test 2 0 200 0

Test 3 20 200 40

Test 4 60 100 0

Test 5 90 130 0

Above table (i.e. Table 2) shows the marks of

student2 in three subjects i.e. subject1, subject2 and

subject3.

Table 3: Marks of Student 3.

Subject 1 Subject 2 Subject 3

Test 1 0 0 100

Test 2 0 100 0

Test 3 100 0 0

Test 4 100 100 0

Test 5 0 100 100

Marks of student3 in different subjects are

presented in the Table 3.

Table 4: Marks of Student 4.

Subject 1 Subject 2 Subject 3

Test 1 150 50 70

Test 2 0 100 0

Test 3 100 0 0

Test 4 100 100 0

Test 5 0 100 100

Marks of student4 in three subjects are presented

in the above table.

C. Methodology, Implementation, and

Experimentation

Now all the numerical data from one text file has to

be represented as one cluster i.e. marks obtained by a

particular student in different subjects for different

tests has to be represented as one cluster. Therefore

four different clusters should be obtained for four

different students i.e. each cluster represents marks

obtained by a particular student.

For showing these clusters in three dimensional

space three axes X-axis, Y-axis and Z-axis are drawn

on a frame developed in Java Language. Each axis

represents a subject. Hence the values in first column

of the text file mapped along the X-Coordinates of a

three dimensional point. Similarly values of second

column are mapped to the Y Coordinates and values

of third column are mapped to the Z-Coordinates. So

marks obtained by a student in three subjects in a

particular test are represented by a point in three

dimensional space.

Numerical data from each file is read and the

values are plotted in three dimensional space using

graphics functions in Java Programming Language.

Mean values for the X-Coordinates, Y-Coordinates

and Z-Coordinates in text file are calculated. The mean

value of X-Coordinates of a particular text file

represents the average value of the marks obtained by

the student in all tests of subject1. This becomes the X-

Coordinate for the data point corresponding to the

mean value of marks obtained by a student in all tests

of three different subjects. In the same way mean

values of Y-Coordinates and Z-Coordinates represents

average marks obtained by a student in all tests of

subject2 and subject3 respectively. These values

become Y-Coordinate and Z-Coordinates for the mean

value data point. It implies that we have a mean value

data point for each student i.e. for each cluster

representing marks of student there is a corresponding

mean value data point. Then a line is drawn from each

data point (representing marks in three subjects) of a

cluster to its corresponding mean value data point.

The lines drawn from each point of a cluster to its

corresponding mean value data give clear picture of

UIVLP: An Improved User Interface and Visualization Technique to Visualize Learners’ Performances

863

deviation of the average marks of all tests from the

marks obtained in each test or looking in a different

perspective we can say that these lines give an idea

about the closeness of the average marks in all tests

from the marks obtained in each test.

Conventionally, in a university setup, for the

relative grading of students’ performance the

visualizations of the frequency-histogram represented

as bar-graph but we propose to use 3D-Visualization

of clusters also to arrive at cutoff for grading because

as we explore later 3D-Visualizations of cluster

provides more insights like depth-queuing and

relative standing of a student’s performance with

respect to three subjects (or three topics) as compared

to the total marks as depicted in the bar-graphs. To

show it we use Blind/Reference less Image Spatial

Quality Evaluator (BRISQUE), no-reference image

quality scores, calculated as per the algorithm given

(Mittal et al., 2012)

, and its interpretation is as

follows, smaller the score better the image quality and

better the visualization considering the image quality.

5 RESULTS

For comparison, in Figs. 1 to 5, we plot bar-graphs

and cluster plots side-by-side and provides specific

details in the figure caption. The following screen

shots presents the output of the Java program

developed as a solution to this problem.

a) BRISQUE

Score = 45.66

b) BRISQUE

Score = 47.635

c) BRISQUE

Score = 44.64

Figure 1: Visualization of marks of student 1, a) Subject

wise totals; subject1 marks are denoted by left-most bar,

subject2 marks are to the right of it, and so on, b) Evaluation

component marks subject-wise; list of bars in the left-most

portions represents marks obtained

y the student in Test1

to Test5 in subject1; to the right of this, subject2 marks fo

five tests taken by the student are displayed and subject3

marks for five tests are displayed right-most, c) 3D display

of marks for five tests clustered about mean-score of three

subjects evaluation-wise for the corresponding student.

In the screen capture (Figure 1) one can clearly

see the graphical visualization in three dimensional

space. Distribution of marks obtained by student 1 is

represented as a cluster. Cluster representing the

marks of student 1 can be clearly seen. Lines are

drawn from each data point to the data point

corresponding to the mean value of the marks of

student 1. Deviation of the average of marks from all

tests from marks obtained in each test in the case of

student 1 can be estimated using these lines.

We observe that because of no slanting lines in

Figure 1 (a) and (b) (i.e. display is mostly generated

by horizontal and vertical lines which are usually

smooth), visually the Figure 1 (a) and (b) may seem

good compared to Figure 1 (c) but image quality

analysis in the terms of BRISQUE score evaluates to

best for the Figure 1 (c).

a) BRISQUE

Score = 46.21

b) BRISQUE

Score = 45.14

c) BRISQUE Score =

44.64

Figure 2: Visualization of marks of student 2, a) Subject-

wise totals; subject1 marks are denoted by left-most bar,

subject2 marks are to the right of it, and so on, b) Evaluation

component marks subject-wise; list of bars in the left-most

portions represents marks obtained by the student in Test1

to Test 5 in subject1; to right of this, subject2 marks fo

fives tests taken by the student are displayed and subject3

marks for five tests are displayed right-most, c) 3D display

of marks for five tests clustered about mean-score of three

subjects evaluation-wise for the corresponding student.

In the above screen capture (Figure 2) one can

clearly see the graphical visualization in three-

dimensional spaces. Distribution of marks obtained

by student 2 is represented as a cluster. Cluster

representing the marks of student 2 can be clearly

seen. Lines are drawn from each data point to the data

point corresponding to the mean value of the marks

of student 2. Deviation of the average of marks from

all tests from marks obtained in each test in the case

of student 2 can be estimated using these lines. Image

quality analysis in the terms of BRISQUE score

evaluates to best for the Figure 2 (c).

ICAART 2025 - 17th International Conference on Agents and Artiﬁcial Intelligence

864

Three dimensional space can be clearly in the

above screen shot (Figure 3) too. Distribution of

marks obtained by student 3 is represented as a

cluster. Cluster representing the marks of student 3

can be clearly seen. Lines are drawn from each data

point to the data point corresponding to the mean

value of the marks of student 3. Deviation of the

average of marks from all tests from marks obtained

in each test in the case of student 3 can be estimated

using these lines. Image quality analysis in the terms

of BRISQUE score evaluates to best for the Figure 3

(c).

a) BRISQUE

Score = 45.51

b) BRISQUE

Score = 48.18

c) BRISQUE

Score = 44.66

Figure 3: Visualization of marks of student 3, a) Subject-

wise totals; subject1 marks are denoted by left-most bar,

subject2 marks are to the right of it, and so on, b) Evaluation

component marks subject-wise; list of bars in the left-most

portions represents marks obtained by the student in Test1

to Test 5 in subject1; to the right of this, subject2 marks for

five tests taken by the student are displayed and subject3

marks for five tests are displayed right-most, c) 3D display

of marks for five tests clustered about mean-score of three

subjects evaluation-wise for the corresponding student.

Figure 4 clearly presents the graphical visualization

of three dimensional spaces. Distribution of marks

obtained by student 4 is represented as a cluster.

Cluster representing the marks of student 4 can be

clearly seen. Lines are drawn from each data point to

the data point corresponding to the mean value of the

marks of student 4. Deviation of the average of marks

from all tests from marks obtained in each test in the

case of student 4 can be estimated using these lines.

Image quality analysis in the terms of BRISQUE

score evaluates to best for the Figure 4 (c).

From Figure 5 (a), we observe that the student1

performs the best, however student1’s performance in

subject3 is inferior as compared to the performance in

subject1 and subject2. This is also evident from the

Table I. The performance of the student3 is poor as

we can see that the entire cluster is closer to the origin.

Similarly, insights about other students can deducted

and if required can be verified from the corresponding

tables.

a) BRISQUE

Score = 47.32

b) BRISQUE

Score = 45.75

c) BRISQUE

Score = 44.64

Figure 4: Visualization of marks of student 4, a) Subject-

wise totals; subject1 marks are denoted by left-most bar,

subject2 marks are to the right of it, and so on, b) Evaluation

component marks subject-wise; list of bars in the left-most

portions represents marks obtained by the student in Test1

to Test 5 in subject1; to the right of this, subject2 marks fo

five tests taken by the student are displayed, and subject3

marks for five tests are displayed right-most, c) 3D display

of marks for five tests clustered about mean-score of three

subjects evaluation-wise for the corresponding student.

If we align the 3D cluster plots of the students

horizontally, vertically or along main diagonal then

we can get more insights. For example, horizontal

arrangement will depict easy understanding of

variations in the performance of the students in

subject2.

The screen-shot, Figure 6, gives complete output

showing the clustering of the data of all the students.

All the four clusters representing marks of all the

students are represented here. This shows

multivariate clustering of the numerical data from all

the files. In a similar way this example can be

extended to a large number of students and clustering

can be visualized as above. Performance of each

student can be assessed by observing this clusters and

distribution of marks gives a clear picture of their

relative performance. Image quality analysis in the

terms of BRISQUE score evaluates to better for the

Figure 6 (b) as compared to Figure 6 (a).

UIVLP: An Improved User Interface and Visualization Technique to Visualize Learners’ Performances

865

a) Marks of Student 1 b) Marks of Student 2

c) Marks of Student 3 d) Marks of Student 4

Figure 5: 3D display of marks for the four students in five

tests clustered about mean-score of three subjects

evaluation-wise.

a) BRISQUE Score =

46.38

b) BRISQUE Score =

44.83

Figure 6: Visualization of marks of the four students, a)

Evaluation component marks subject-wise, b) 3D display o

marks clustered about mean-score evaluation-wise.

6 CONCLUSIONS

The problem of visualization of students’

performance in terms of marks obtained in various

subjects can be solved using the Cluster Analysis

technique of multivariate analysis. Graphical

visualization of clusters representing the marks of

multiple students in different subjects is implemented

in Java. The distribution of marks of the students and

deviation from average of marks from all tests from

marks obtained in each test can be clearly observed

from the graphical visualization provided. The 3D

visualizations are compared with the bar-graph using

no-reference image quality scores, BRISQUE, and it

is observed that the BRISQUE scores are slightly

better than the 2D bar-graph displays because lower

the BRISQUE score better the image quality. In

addition to this, it is observed that 3D cluster plots

provide 3D clues and depth cues for further

differentiating the students’ performance for grading

by considering marks in three subjects rather than just

the total marks. As a future work, user interface (UI)

can be developed for arranging the 3D plots along a

selected axis so that UI can provide more insights.

These insights can help in deciding the boundary

cases for adjusting the cut-off for grading. In addition

of this, future use of k-means clustering or Isodata

algorithms can be made to visualize the students’

performance along selected or most discriminating

dimensions.

REFERENCES

Hearn, D., & Baker, M. P. (2015). “Computer Graphics C

Version”, Pearson Education, 2003.

Johnson, R. A., & Wichern, D. W. (2007). Applied

multivariate statistical analysis, Pearson Education Inc.

Al-Barrak, M. A., & Al-Razgan, M. (2016). Predicting

students final GPA using decision trees: a case

study. International journal of information and

education technology, 6(7), 528.

Misailidis, E., Charitopoulos, A., & Rangoussi, M. (2018,

November). Visualization of educational data mined

from the moodle e-learning platform. In Proceedings of

the 22nd Pan-Hellenic Conference on Informatics (pp.

82-87).

Humphries, W., Gawrilow, J., Turner, S., Perez-Quinones,

M., & Edwards, S. (2006, June). Helping Students

Visualize Their Grade Performance. In 2006 Annual

Conference & Exposition (pp. 11-683).

Deng, H., Wang, X., Guo, Z., Decker, A., Duan, X., Wang,

C., ... & Abbott, K. (2019). PerformanceVis: Visual

analytics of student performance data from an

introductory chemistry course. Visual Informatics, 3

(4), 166–176.

Darcy Dario (2022), 12 Learning Analytics Software,

Published October 6, 2022, EdApp LMS Homepage,

ST Training, https://training.s afetyculture.com/

blog/learning-analytics-software/

Paolucci, C., Vancini, S., Bex II, R. T., Cavanaugh, C.,

Salama, C., & de Araujo, Z. (2024). A review of

learning analytics opportunities and challenges for K-

12 education. Heliyon.

ICAART 2025 - 17th International Conference on Agents and Artiﬁcial Intelligence

866

Mukred, M., Asma’Mokhtar, U., Hawash, B., AlSalman,

H., & Zohaib, M. (2024). The adoption and use of

learning analytics tools to improve decision making in

higher learning institutions: An extension of technology

acceptance model. Heliyon, 10(4).

Atif, A., Richards, D., Bilgin, A., & Marrone, M. (2013).

Learning analytics in higher education: a summary of

tools and approaches. In Ascilite-australian society for

computers in learning in tertiary education annual

conference (pp. 68-72). Australasian Society for

Computers in Learning in Tertiary Education.

Mittal, A., Moorthy, A. K., & Bovik, A. C. (2012). No-

reference image quality assessment in the spatial

domain. IEEE Transactions on image

processing, 21(12), 4695-4708.

UIVLP: An Improved User Interface and Visualization Technique to Visualize Learners’ Performances

867