BUSINESS INTELLIGENCE IN HIGHER EDUCATION

Managing the Relationships with Students

Maria Beatriz Piedade

School of Technology and Management, Polytechnic Institute of Leiria, 2411-901 Leiria, Portugal

Maribel Yasmina Santos

Information Systems Department, Algoritmi Research Centre, University of Minho, 4800-058 Guimarães, Portugal

Keywords: Business Intelligence, Customer Relationship Management, Data Mining, Data Warehouse, OLAP, Student

Relationship Management.

Abstract: The closely monitoring of the students’ academic activities, the evaluation of their academic success and the

approximation to their day-by-day academic activities are key factors in the promotion of the student’s

academic success in higher education institutions. To be possible the implementation of monitoring

processes and activities, it is essential the acquisition of knowledge about the students and their academic

behaviour. This knowledge supports the decision-making associated with teaching-learning process,

enhancing an effective institution-student relationship. This paper presents a Student Relationship

Management (SRM) system that is under development. The SRM system supports the SRM concept and

practice and has been implemented using concepts and technologies associated to the Business Intelligence

systems. To demonstrate the SRM system relevance in the process of acquisition of knowledge about the

students and in the support of actions and decisions based on such knowledge, an application case carried

out in a real context is also presented.

1 INTRODUCTION

Portuguese Higher Education has been characterized

by a high rate of students’ failure and abandon,

mainly in the first year of the graduation courses

(statistical results can be consulted in the official

web page http://www.gpeari.mctes.pt). Although this

reality has changed due to several individual or

institutional actions, integrated activities need to be

proposed and adopted. To be possible, it is necessary

the identification of the factors and the measures that

need to be monitored in the teaching-learning

process and in the student-teacher relationship. One

of the activities usually pointed out as crucial to

promote the students’ success is the closely

monitoring of the students’ academic activities (Pile

and Gonçalves, 2007). Although important, this

activity does not take place in many institutions.

Among the reasons, we point out the huge number

of students with failure in the first graduation year,

the huge number of new students in some courses

and the work overload of the teaching staff. In the

Portuguese institutions, teachers are involved in

lecturing, researching and management tasks. To

help teachers and students in this complex process,

an adequate conceptual and technological support is

needed. The conceptual framework and the

technological infrastructure are in this work

integrated in a Student Relationship Management

(SRM) system. The SRM system supports the SRM

concept and practice and is based in a Business

Intelligence infrastructure. To demonstrate its

relevance in the students’ knowledge acquisition

process, in the early identification of failure

situations, in the decision-making support and in the

automatic interaction with the students, it is

presented an application case that has occurred in a

real context. This paper is organized as follows:

Section 1 refers the academic failure as the problem

that motivates and justifies the SRM system; Section

2 includes an overview of the SRM principles and

concepts; Section 3 describes the SRM system

architecture; Section 4 describes the application case

and the data analysis process. This section also

297

Beatriz Piedade M. and Yasmina Santos M. (2009).

BUSINESS INTELLIGENCE IN HIGHER EDUCATION - Managing the Relationships with Students.

In Proceedings of the International Conference on Knowledge Discovery and Information Retrieval, pages 297-302

DOI: 10.5220/0002302602970302

 SciTePress

presents the interpretation of the obtained results and

their support to interaction activities with the

students; Section 5 concludes with the SRM system

expected advantages and the upcoming future work.

2 CONCEPTUAL FRAMEWORK

The CRM (Customer Relationship Management)

systems are nowadays used in business contexts to

support and to manage the relationships with the

customers. These systems allow the identification of

knowledge about the customers, using the

information and business transactions available in

the organization databases. Using this knowledge,

the organization defines the activities and actions

that allow maintaining a close and strong

relationship with its customers (Payne, 2006). The

SRM system was inspired in the CRM principles, but

supports processes and activities concerned with the

teaching-learning process, mainly activities that

allow the monitoring and the supervision of the

students academic activities. To exemplify the

similarity between the CRM/SRM actions, one could

compare the actions developed by the customer’s

manager, that on the scope of the banking activity,

alerts the customer when he/she exceeds his/her

credit account, and the actions developed by the

student’s tutor/teacher, that on the scope of the

monitoring processes send an alert message to the

student when detects he/she misses several lessons.

To refer, that the “Student Relationship

Management” designation was already used in a

technological/commercial environment to designate

solutions mainly dedicated to support processes

related with the students in academic areas

(students’ management information, courses and

lessons management, admissions management,

enrolment and registration management) and areas

related with available services (communications,

marketing, financial aids, accommodation). In this

work is proposed a definition of the SRM concept,

understood as a process based on the students’

acquired knowledge, whose main purpose is to keep

an effective student-institution relationship through

the closely monitoring of the students’ academic

activities. This concept was based on the premise

that there exist a strong correlation between the

closely monitoring of the students’ academic

activities and their academic success promotion. The

SRM practice is understood as a set of

activities/actions, which should guarantee the

students’ individual contact, and an effective,

adequate and closely monitoring of his/her academic

performance. To validate the SRM concept and the

set of activities included in the SRM practice it was

adopted a methodology based on the Grounded

Theory principles, which included the interviews

realization and analysis (Piedade and Santos, 2008).

3 SRM SYSTEM

The SRM system is based on the SRM concept and

on a set of activities that composes the SRM

practice. To undertake a SRM practice it is

necessary: (i) to have adequate, consistent and

complete information about the students. This

information must be stored in an appropriate data

repository, which allows maintaining a single vision

of students’ data; (ii) the analysis of such data in

order to obtain knowledge about the students and

their academic behaviour; (iii) the starting of

automatic actions whenever specific situations are

detected; and (iv) to assess the impact of all the

implemented actions over the students. With respect

to i) and ii) the SRM system envisages the

implementation of a data warehouse and its

exploration using data analysis tools. These

structural framework leads that the SRM system is

implemented using the technological infrastructure

that traditionally supports the Business Intelligence

systems (Negash and Gray, 2003). With respect to

iii) and iv) it was identified the relevant indicators

and behaviour patterns that characterizes the

different situations to supervised and it was

implemented the actions to be executed

automatically by the different participants in the

SRM practice (teacher, tutor, course director) and

analysed the impact of the carried out actions in the

students behaviour and their final results.

The SRM system architecture aggregates four

main components: (i) The Data Acquisition and

Storage component, responsible for the storage, in

the data warehouse, of the students’ data; (ii) the

Data Analysis component, responsible for obtaining

knowledge about the students, using appropriate data

analysis tools that allows the patterns identification;

(iii) the Interaction component, responsible for

maintaining an adequate and effective relationship

with the student. In this interaction is used the

knowledge about the student(s) to start a set of

automatic actions that reflect the academic situation

of an individual or a set of individuals; (iv) the

Assessment component, responsible for the

assessment of all the concretized actions and their

impact. The SRM system prototype implementation

was done using database management tools,

KDIR 2009 - International Conference on Knowledge Discovery and Information Retrieval

298

Business Intelligence tools and web development

tools. Considering the context in which this project

takes place, the selected development tools integrate

the Microsoft environment. The SRM system

prototype validation has been done through the

execution of a set of application cases, taking place

in different Higher Education Institutions (Piedade

and Santos, 2009).

4 APPLICATION CASE

The data used in this application case was gathered

in a graduate course of a Portuguese Higher

Education Institution. This course, in the

engineering area, is composed by a set of curricular

units. The selected unit, with 70 students, is

integrated in the first graduation year. The teaching

method adopted is based in a presential component,

and on an e-learning component supported by the e-

learning platform available in the institution. The

presential component includes different types of

classes (theoretical, practical and tutorial

orientation). The activities included in the theoretical

classes are related with the curricular subjects’

presentation and explanation. The activities included

in the practical classes are related with exercises and

problems solving in a laboratory environment. The

activities included in the tutorial orientation classes

include students’ individual work support and

orientation and subject clarification. The e-learning

component includes activities related with the

distribution of relevant information and materials

(unit general information, curricular contents,

exercises and project guidelines), and, also,

communication activities (messages that are

automatically or manually sent and discussion

forums). The unit assessment includes two distinct

methods: the normal assessment period and the

exam assessment period. The normal assessment

period integrates a written test and a project with

individual discussion, with weights of 40% and

60%, respectively, in the final mark. The

quantitative mark scale comprises values among 0 to

20. In both assessments, 8 is the minimum mark that

needs to be obtained by the students in order to be

possible the calculus of their final mark. This final

mark results from the application of the weights

associated with the test and the project and needs to

be equal or higher than 10 to guarantee success in

the unit. The exam assessment period only

comprises a written exam. To pass the unit, the

student must obtain a positive mark (value greater or

equal then 10). In the exam evaluation, we

frequently have students who have failed the normal

assessment and/or students that missed the normal

assessment. The available data, about each student

and his/her involvement in the teaching-learning

process, was: (i) provided by the institutional

academic system (like students’ personal

information and unit information); (ii) provided by

the unit teachers (include students presences in

classes, developed activities and the corresponding

student marks); (iii) provided by the e-learning

system (information related with the student-unit

interaction using the e-learning platform).

The analysis of all the available data allowed the

identification of the data subset considered in this

application case. This data subset includes:

 student number; student registration year; and

student phase of admission to higher education. The

phase information is only related with first year

students and can have the values first or second.

This attribute is also used to identify the students

that are repeating the unit (value rep), as a

consequence of a previous failure; and, worker/full

time students information. In order to maintain the

students’ privacy, all the information that allows

his/her identification is ignored or codified.

 unit identification; unit designation; unit

curricular year and semester; and associated course.

 class type identification; class type description;

class start hour; class duration; and, class week day.

 assiduity rate associated with each student and

each class type. The assiduities values were

transformed in the following classes:

Low (< 50%),

Acceptable (≥ 50 and < 70%) and High (≥ 70% and

≤ 100%).

 assessment activity identification; activity

description; weight in the final mark, mandatory

information; and, marks (obtained by each student).

To represent some specific situations, negative

values were used. In the project assessment results

the value -1 mean that the student misses the project

individual discussion; the value -3 mean that the

student did not implemented the project work. In the

unit final results, the value -1 mean that the student

was not submitted to any type of evaluation (test,

project or exam); -2 mean that the student failed the

unit, but he/she was submitted to any one of the

activities includes in the unit assessment (test/

project/exam). The final marks were also classified

in qualitative terms, using for that purpose the

following attributes: Satisfactory (between 10 and

13), Good (between 14 and 16) and VeryGood

(between 17 and 20).

 number of distinct days that each student

interacted with the unit using the e-learning

BUSINESS INTELLIGENCE IN HIGHER EDUCATION - Managing the Relationships with Students

299

platform. In qualitative terms, it was considered that

from 0 to 16 corresponds to a low interaction, 17 to

32 corresponds to a reasonable interaction, 33 to 49

corresponds to an expressive interaction and values

greater or equal to 50 correspond to a high

interaction (the distributions of the values was

analysed in order to be possible the definition of this

limits).

Considering all the relevant data, it was: (i)

designed the data warehouse model, a

multidimensional data model which follows the

constellation schema (Figure 1); (ii) implemented

the data warehouse, by fact and dimension tables

creation; (iii) loaded the operational data to the data

warehouse. The loading process followed the ETL

process steps, in which the relevant data was

extracted from the source databases, was cleaned

(when errors in data were detected) and was

transformed in order to accomplish the format of the

target system (the data warehouse).

Figure 1: The Data warehouse model.

The data warehouse exploration has been done

using OLAP and data mining techniques.

In this application case, OLAP cubes were

created to analyse the students’ results verifying

both the teaching-learning experiences and the

assessment methods influence. To analyze

particularly the unit results, and the correlation with

the theoretical classes’ presences, the unit interaction

through the e-learning platform and the project work

a cube was created. From its analysis it is clear that

all the students that did not implement the project

(mark value -3) or misses the individual project

discussion (mark value -1) fail the unit. It can also

be verified that there exists a large number of

students with low assiduity to theoretical classes.

Many of them are repeating the unit and others are

differentiated by the phase of admission to the

University. In these cases, only the students that are

repeating the unit pass. These students also have a

reasonable or an expressive interaction with the e-

learning platform. The students that fail have few

interactions with the unit, existing only one

exception, a student (id 15) with many interactions

(Figure 2).

Figure 2: Students data extract, grouped by Low assiduity

rate.

Another analysis (Figure 3) allows us to verify that

many students with high assiduity rates, and

expressive or many interactions, pass the unit with

good marks. Students with acceptable assiduities

and few or reasonable interactions, fail or pass the

unit with a sufficient mark. Preoccupant situations

occur with students that go to the University in the

second phase, as many of them fail the unit. It was

verified that some students fail and others pass the

unit with different marks.

Figure 3: Students data extract, grouped by Acceptable

and High assiduity rate.

Now our objective is to identify the behaviour of

the students differentiating them by marks. The main

KDIR 2009 - International Conference on Knowledge Discovery and Information Retrieval

300

purpose is to identify the students’ profile. Through

it, mainly in the case of fail, adequate actions can be

implemented to minimize the failure rate in future

unit editions.

Data mining algorithms were used to identify the

students’ profile considering the assessment results:

Fail, Satisfactory, Good, Very Good. The objective

is to find a model that describes the predictable

attribute, Mark, as a function of the input attributes

phase, situation, numDays and the assiduity rates for

the theoretical, practical and tutorial classes. It was

followed the traditional steps of the knowledge

discovery in databases process: Data Selection, Data

Treatment, Data Pre-Processing, Data Mining and

Results Interpretation. The Data Selection, Data

Treatment, Data Pre-Processing steps were

supported by the data warehouse implementation

process. In the Data Mining step, it was selected a

decision tree algorithm to carry out the classification

task previously defined. The obtained model (Figure

4) integrates a set of rules, in a tree form, where each

tree node has associated a set of conditions that lead

to a specific decision.

All

theoretical=’low’

theoretical=’high’

theoreti cal=

’acceptable’

numDays not

=’expressive’

numDays=

’expressive’

numDays not=

’reasonable’

numDays=

’reasonable’

phase=

’rep’

phase=’1'

Phase=’2'

practical=

’high’

practical not=

’high’

numDays not

=‘many’

numDays

=’many’

practical=

’high’

practi cal not=

’high’

Fail

Satisfactory

Good

Phase not=

’rep’

phase=’rep’

Fail

Satisfactory

Good

Sati sfactory

phase=’2'

phase not =’2'

Fail

Figure 4: Data mining model.

The identified model integrates a set of rules that

explicitly describe the students’ profile. From the

analysis of the rules, it is possible to verify that the

theoretical attribute is the attribute that more

influence the students’ profile, followed by the

phase, practical and numDays attributes.

The Fail profile is associated with:

1. Repeating students with low theoretical assiduity

rates and few, reasonable or many interaction

with the e-learning platform;

2. Students at the first time in the unit, with low

theoretical assiduity and an expressive

interaction level;

3. Students with the first registration (first phase),

with an acceptable theoretical assiduity and with

few or acceptable practical assiduity;

4. Students with the first registration (second

phase) with an acceptable theoretical assiduity.

The Satisfactory profile is associated with:

1. Repeating students with acceptable theoretical

assiduities, or low theoretical assiduities, but

with an expressive interaction;

2. Students with the first registration in the unit,

with acceptable theoretical assiduities and high

practical assiduities

The Good profile is associated with:

1. Students with high theoretical assiduities.

No rule was obtained to characterize the Very Good

profile. This is due to the fact that only one student

achieved this mark.

The OLAP and data mining analyses allow us to

verify that to decrease the failure profile it is

necessary to take special attention to the students,

differentiating them by the admission phase.

The students that are at the first time in the unit

must be encouraged to go in a regular basis to the

different type of classes. These students are, in many

cases, influenced by older students that say to them

to avoid classes, mainly the theoretical classes. For

the second phase students, an additional support

must be given as they arrive to the University when

half semester has passed. Due to this situation, these

students lose the initial curricular contents

explanation and the subsequent curricular content

comprehension, fact that could help to explain their

failure. For these cases, the institution could adopt

special activities or procedures, as extra classes or

tutorial orientation, providing additional support to

the students. It is also necessary, for all the students,

to verify the evolution of the project work

implementation, motivating the students and

providing additional support when necessary. This

support can be achieved using tutorial classes. This

was not the case of the current edition of the unit,

helping to explain the lack of presences and interest

in this king of lectures. In what concerns repeating

students, it is also necessary to motivate them to go

to the presential classes, although in many situations

these students have timetable incompatibilities. To

overcome this limitation, the institution could adopt

the schedule differentiation, like classes in the

morning for the first year students and in the

afternoon for the second year students. In

complement, it is also necessary to motivate the use

of e-learning

platform increasing the interaction of

the student with the unit. These different situations

need to be evaluated without neglecting the students

that present what we could say is a “success profile”.

The previous results allow us to conclude that

data analysis supports the process of obtaining

knowledge about the students and their academic

BUSINESS INTELLIGENCE IN HIGHER EDUCATION - Managing the Relationships with Students

301

behaviour, through the students’ behaviour pattern

identification.

The obtained students’ information allows the

definition of a set of actions (included in the SRM

practice) to closely follow the students. The main

purpose is to decrease the failure rate. The identified

actions, to integrate in future editions of the unit,

include the presences/interaction automatic

monitoring and an adequate support to the project

work. The presences/interaction automatic

monitoring needs the automatic recording of this

data and the subsequent automatic sent of alert

messages to the student when deviating behaviours

are detected. The main purpose of these actions is to

alert the students to both the continuous unit

interaction and the daily study as a way of increase

their academic success. Between the teachers and

the course coordinator/director could also exist

periodically changes of information, which state

how the semester is ongoing. Based on these reports

specific actions can be took, as verifying if the

student has any problem.

The implementation of activities, like described

above, is always supported by a web application.

Figure 5 shows an example of the presences report.

To refer that the descriptions are in the Portuguese

language and neither the student nor the institution is

identified.

Figure 5: Application web page view.

Next steps, in this project include: i) the

integration of additional information in the data

analysis process, related with the students and their

activities in the unit scope ii) the implementation of

other activities that enable a closely monitoring of

the students (activities included in the SRM

practice); and (iii) the SRM practice assess, through

the analysis of its impact in the students behaviour

and their final results.

5 CONCLUSIONS AND FUTURE

WORK

In Portuguese Higher Education institutions exists a

strong budget control and persists a high rate of

failure and abandon (mainly in the first graduation

year). With the new formative process

implementation aligned with the Bologna Process,

the number of hours of contact between the teachers

and the students decreased. This requires high

student autonomy in the learning process. In this

scenario, it is essential the design and

implementation of mechanisms that facilitate the

monitoring of the students’ academic activities. In

this context, we believe that the SRM concept and

practice implementation, supported by the SRM

system, creates an advantage towards the students

success promotion, and, therefore, in the institution

success, ensuring an effective student-institution

relationship. Future work, in this project, includes

the fully implementation of the prototype and its

validation with more application cases (running in

two higher education institutions).

REFERENCES

Negash, S. and Gray, P. 2003. "Business Intelligence." In

Ninth Americas Conference on Information Systems.

Payne, A. 2006. Handbook of CRM. Achieving

Excellence in Customer Management. Elsevier-BH

Piedade, M. B. and Santos, M. Y. 2008. "Student

Relationship Management: Concept, Practice and

Technological Support." In IEMC 2008 - IEEE

International Engineering Management Conference.

June 28 to 30, Estoril, Portugal.

Piedade, M. B. and Santos, M. Y. 2009. "An Application

of the Student Relationship Management Concept." In

ICEIS 2009 - 11th International Conference on

Enterprise Information Systems. May 6 to 10, Milan,

Italy.

Pile, M. and Gonçalves, I. 2007. "Programa de

Monitorização e Tutorado." In A qualidade em

estabelecimentos de ensino superior. Exemplo de boas

práticas. Instituto Português da Qualidade. Caparica.

KDIR 2009 - International Conference on Knowledge Discovery and Information Retrieval

302