A Decision-Making Approach Combining Process Mining, Data

Mining and Business Intelligence

Olfa Haj Ayed

and Sonia Ayachi Ghannouchi

1,2 b

Higher Institute of Management of Sousse, University of Sousse, Tunisia

Laboratory RIADI-GDL, ENSI, University of Mannouba, Tunisia

Keywords: Process Mining, Data Mining, Business Intelligence, Visualisations, Dashboards.

Abstract: In the era of Big Data, Process Mining (PM), Data Mining (DM) and Business Intelligence (BI) are essential

analytical tools for companies. By intelligently exploiting big data, these approaches make it possible to

extract valuable information. Although each has its own orientation, concepts, techniques and modes of

visualization, these three disciplines converge towards a common goal: improving decision-making. This

work proposes an innovative approach which consists in combining the strengths of PM, DM and BI within

a powerful global dashboard. This centralized dashboard will bring together visualizations from all three

domains, providing a holistic and interactive overview of key business data. By providing decision-makers

with these information-rich visualizations, the study aims to facilitate and accelerate the decision-making

process, thus allowing informed and responsive strategic choices.

1 INTRODUCTION

In today's information age, informed decision-making

has become a critical issue for businesses. To achieve

this, decision-makers need a comprehensive

overview and a deep understanding of the company's

data. A dashboard, bringing together visualizations

from Process Mining, Data Mining and Business

Intelligence, becomes a valuable tool to meet this

need.

A dashboard combining PM, DM and BI provides

many benefits to decision makers such as having a

global vision, it offers a synthetic overview of the

company by grouping information from different

sources and disciplines. The visualizations from PM,

DM and BI allow to analyse the data in depth, to

identify trends, anomalies and opportunities. By

having a global view and a deep understanding of

data, decision makers can make more informed and

strategic decisions. In addition to effective problem

solving, the dashboard facilitates the identification of

problems and the implementation of adequate

solutions. Visualizations make it possible to follow

the evolution of the key performance indicators

(KPIs) and to measure the effectiveness of the

https://orcid.org/0009-0004-5610-0817

https://orcid.org/0000-0001-9583-9797

processes. Finally, having a performance

improvement and a better decision-making process

contribute to the improvement of the overall

performance of the company. Our work is part of this

approach by proposing the creation of a powerful

global dashboard combining the strengths of PM, DM

and BI. The major contributions of our work are:

 Creation of intermediate PM, DM and BI

dashboards which allow to structure and

analyse the data from each domain

independently.

 Development of a powerful global dashboard

which centralizes the visualizations of the three

domains, offering a coherent and interactive

overview.

The rest of this paper presents in detail the

different stages of our work: Section 2 presents our

background. Regarding the third section, it deals with

the related work. In section 4, we illustrate the

research problem and our proposed approach. Section

5 presents the guidance tool. Section 6 presents our

case study result. In section 7, we conclude this paper

450

Haj Ayed, O. and Ghannouchi, S.

A Decision-Making Approach Combining Process Mining, Data Mining and Business Intelligence.

DOI: 10.5220/0012852500003753

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

In Proceedings of the 19th International Conference on Software Technologies (ICSOFT 2024), pages 450-457

ISBN: 978-989-758-706-1; ISSN: 2184-2833

2 BACKGROUND

Process mining (PM) is defined as a technology that

utilizes event logs corresponding to real Behavior

recorded during the execution of a business process.

It helps to discover, monitor, and improve processes

in real time by extracting knowledge available in

system log files. It leads to delivering an assessment

on the conformance status of business process

execution (Van der Aalst, 2016).

Data Mining (DM) is defined as a process that

aims to generate knowledge about very large

databases and to produce results in a comprehensive

way to the user. Indeed, DM extracts non-trivial,

implicit, previously unknown and potentially useful

information (Schuh et al., 2019).

Business Intelligence (BI) is a set of methods,

processes, architectures, applications, and

technologies that collect and transform raw data into

meaningful and useful information used to enable

strategic, tactical, and operational insights and more

effective decision-making to drive business

performance (Tripathi et al., 2020).

Data Visualization is the graphical representation

of information extracted from raw data. It consists of

transforming complex and abstract data into images,

tables, graphs and other visual elements that are easy

to understand and interpret. The goal is to make data

easier to understand by making it more accessible and

intuitive (Azzam et al., 2013).

3 RELATED WORK

The scientific literature is full of relevant work

exploring the use of dashboards in BI (Orlovskyi &

Kopp, 2020), PM (Martinez-Millana et al., 2019), and

DM (Maya D. Albayrak & William Gray-Roncal,

2019) approaches. This work demonstrates the

usefulness of dashboards to visualize and analyse data

from different sources, thus facilitating informed

decision-making. This section presents an overview

of relevant previous work related to combination of

approaches.

In (Kumar SM & Meena Belwal, 2017), the

authors use BI, DM and data visualization

technologies to create a scoreboard that presents the

information by underestimating the behaviour of the

company from its inception. In addition, it provides

an overview to users, making complex datasets easier

for them to use, and it also tracks the ability of the

service to meet service level objectives. Based on

several recent works the researchers were able to

create a powerful Dashboard by adding more features

to what is already created among these new functions

including the integration of BI technologies, Data

mining and data visualization technologies to analyse

business trends, business growth, profit amount,

employee performance, customer satisfaction and

improvement areas. The proposed performance

dashboard features an ideal single-pane real-time user

interface, showing a graphical presentation of the

historical status and trends of organizations' key

performance indicators that enable executive

decision-making at a glance and improve business

performance.

In (Nik et al., 2019), the authors describe a custom

visual of Microsoft Power BI, called BIpm, which

was created by combining Process Mining and

Business Intelligence Analysis through a single

platform. To achieve their objectives the researchers

went through several steps, starting with the

preparation of the input fields and placing them in the

Power BI pane as well as the event logs, Let’s not

forget that Process Mining is a technology that

requires the presence of event logs to determine the

behaviour of processes so it is necessary to have

events logs consistent with Power BI. Once all these

fields are entered correctly, BIpm creates the process

model as a directed flow graph. BIpm offers an online

analysis for decision makers in industry. This solution

allows to analyse complex events logs, on the one

hand it enriches the BI dashboards with the

exploration of interactive online processes, and on the

other hand it allows BI users to expand their toolsets

by inferring process models.

According to (Hendricks, 2019), DM can be

used in the field of health, but not only the DM, there

is also the PM which seems similar to the DM in

terms of measuring large data files, but in this case,

we are talking about event logs to a particular process

or a series of processes. The PM was performed on a

Dutch patient hospital log event with sepsis entering

the emergency room, to understand this method of

analysis, highlight the information discovered and

determine its role in data mining, and their release and

possible readmission stages. This analysis makes it

possible to map and analyse the processes, and also to

highlight the areas of clinical operations requiring

further investigation including a possible relationship

with the patient’s readmission and method of release.

A Decision-Making Approach Combining Process Mining, Data Mining and Business Intelligence

451

4 RESEARCH PROBLEM AND

PROPOSED SOLUTION

This section discusses the research problem and our

proposed solution.

4.1 Research Problem

Effective decision-making relies on a comprehensive

understanding of the entire business environment and

granular data. To facilitate this task, dashboards

provide a centralized space for data visualization and

activity tracking, enabling informed decision making

and effective problem solving.

Technological developments, particularly in the

field of information technology, have revolutionized

the decision-making process. Well-designed and

organized dashboards enable decision makers to

efficiently navigate large amounts of data, turning

raw information into actionable insights. The

importance of sound decision-making cannot be

overstated, as it has the potential to propel an

organization to new heights or to its downfalls.

Recognizing the strengths of each individual

approach, our research focuses on integrating BI, DM

and PM methodologies to create a powerful

dashboard that empowers decision-makers.

Specifically, we address the following research

questions:

 How much is it useful to combine the three

approaches of DM, PM and BI?

 How to produce a dashboard for a decision-

maker considering the three approaches DM,

PM and BI?

4.2 Proposed Solution

Our research aims to develop a powerful dashboard

that facilitates informed decision-making for

managers by combining the strengths of PM (Project

Management), BI (Business Intelligence) and DM

(Data Mining) approaches. To do this, we propose the

creation of intermediate dashboards specific to each

approach, followed by the integration of the most

relevant visualizations in a powerful global

dashboard.

 Intermediate BI Dashboard: This dashboard

will focus on analyzing and visualizing

business data, providing decision makers with

an overview of key performance indicators

(KPIs) and business trends.

 Intermediate DM Dashboard: This dashboard

will focus on exploring and analyzing

operational data, enabling decision makers to

identify opportunities for process and decision-

making improvement.

 Intermediate PM Dashboard: This dashboard

will focus on project management and task

tracking, providing decision makers with

visibility into project progress and potential

risks

Figure 1 : Overview of the proposed approach.

The judicious selection and combination of the most

relevant and effective visualizations from the

intermediate dashboards will be essential to create a

powerful global dashboard. This global dashboard

will provide decision-makers with a comprehensive

and synthetic view of key information for decision-

making.

To achieve our main objective, it is crucial to define

precise and repetitive sub-objectives for each

approach (PM, BI and DM). These sub-objectives

will guide the selection of the data to be analysed, the

choice of the types of visualization to adopt and the

way to present the final dashboard.

Our approach is distinguished by its ability to

facilitate decision-making and providing managers

with clear and relevant visualizations from the

intermediate dashboards. The advantage lies in the

combination of the three major approaches that are

ICSOFT 2024 - 19th International Conference on Software Technologies

452

PM, DM and BI, thus offering a global and informed

perspective for decision-making.

As shown in Figure 1, we propose to divide this

part into two sub-parts, the first will be dedicated to

the steps followed to develop the three intermediate

dashboards. While the second part will be devoted to

the creation of our global dashboard that combines

visualisations of intermediate dashboards.

Our work involves close collaboration between

business experts, end users of dashboards, and

analysts, in charge of their design. The data comes

from an online sales website. The creation process

goes through several stages. The analyst selects and

prepares (cleans) the relevant data for DM, PM and

BI analyses, guaranteeing their quality and reliability.

Subsequently, in collaboration with the business

expert, the analyst offers a guidance tool that defines

the types of visualizations most suited to the three

approaches and it is up to the trade expert to choose

the visualisations. Then, based on the jointly defined

visualization choices, the analyst selects an

appropriate methodology to create the intermediate

dashboards (PM, DM and BI), making full use of the

prepared data. Finally, the analyst combines

intermediate dashboards, ensuring optimal

consistency and fluidity of information, to create an

informative and intuitive global dashboard.

This collaborative approach ensures that

dashboards meet the specific needs of business

experts while leveraging analysts' expertise in data

processing and visualization. The result is a powerful

dashboard that facilitates informed decision-making

and promotes the achievement of strategic objectives.

5 GUIDANCE TOOL

The guidance tool, as shown in figures 2 and 3,

developed with the Angular framework, offers an

intuitive and user-friendly interface consisting of four

sub-interfaces dedicated to the selection of

visualizations for BI, DM, PM and the global

dashboard. Each sub-interface presents a list of

relevant visualizations accompanied by an "eye" icon.

By clicking on this icon, the business line manager

accesses a detailed description of the type of

visualization and an illustrative example. To select

the desired visualizations, the business manager

simply ticks the corresponding boxes and clicks on

the "Export" button. An Excel file is then generated,

containing the complete list of selected visualizations.

The business line manager then sends the Excel file

to the analyst, who uses it as the basis for creating the

intermediate dashboards and the global dashboard.

This collaboration ensures that dashboards meet the

specific needs of the business leader while leveraging

the analyst’s expertise in data processing and

visualization.

Figure 2: Overview of the guidance tool.

Figure 3: Interface of BI dashboard construction in the

guidance tool.

6 CASE STUDY

Our case study focuses on the field of e-commerce,

focusing more specifically on the online sales of

medical products. To carry out this analysis, we have

exploited an extensive database from an e-commerce

website specialized in this field. This information-

rich database allowed us to explore customers' buying

behaviours, identify market trends and draw valuable

conclusions for optimizing online sales strategies for

medical products.

6.1 Proposed Solution

For the development of the three intermediate

dashboards, we follow a three-step process: data

selection and preprocessing, visualization technique

selection, and dashboard design.

6.1.1 Choosing and Preprocessing Data to

Analyse

Choosing the right data for analysis, whether it’s BI,

PM or DM, is crucial to making informed decisions

and extracting valuable information from the

A Decision-Making Approach Combining Process Mining, Data Mining and Business Intelligence

453

organization’s information assets. This fundamental

step is to understand the content of the data sources

and carefully align the data with the specific

objectives of the analysis. Let’s take the example of

Process Mining (PM), which aims to analyse business

processes from system logs. To analyse a document

with the PM, it is crucial to have a log file containing

at least three mandatory fields, Case ID, Activity and

Timestamp.

Rigorous data selection and preparation is a

fundamental step for effective analysis, whether in

BI, PM or DM. By following these key steps, we

ensure that the data used is relevant, reliable and

aligned with the objectives of the analysis, thus

obtaining valuable and actionable information.

6.1.2 Choosing Visualisation

We detail below which visualisation is chosen for

each domain.

Choosing Visualisation for DM:

The decision tree is a powerful tool for classifying

product usage patterns (Breşfelean, 2007). To classify

the usage patterns of medical products according to

various factors, the expert chose the decision tree, a

method known for its effectiveness and simplicity of

interpretation. Three separate decision trees were

constructed to explore specific aspects of usage

patterns. The first decision tree focused on gender.

Whereas, the second tree, focused on age category.

The last decision tree revolved around season, which

is considered the most important, as it analyses the

impact of the seasons on the trends of use of the

products. The analysis shows that the use of products

is more intensive in spring, while it becomes punctual

in summer, autumn and winter. In addition, product

sales are higher in Spring than in Summer, Autumn

and Winter.

Figure 4: Customers' buying habits according to the season.

These insights allow the decision maker to adapt its

marketing strategy according to the seasons by

proposing relevant products to meet the needs of

customers at each period of the year (fig 4)

The use of SimpleKMeans algorithm might be

useful. It’s a clustering technique recognized in DM.

The use of Scatter Plot in DM is particularly

interesting for its simplicity and flexibility (Soma

Ajibade & Adediran, 2016). It allows an easy and fast

understanding of the data by the decision-makers. In

this case, Scatter Plot illustrates the number of orders

placed by age groups.

Choosing Visualisation for BI:

As part of this analysis, the expert selects five key

types of visualizations to effectively communicate

critical information which are maps, histogram, bar

graph, curve diagram and Pie Chart.

Figure 5 : The greatest product.

Histograms present a classic and effective method for

comparing and classifying multiple elements (Airinei

& Homocianu, 2010), In our case, this visualization

proved particularly useful to identify the best-selling

product. The analysis of the histogram allowed to

deduce the product which knows the greatest success

with the customers (fig 5). The expert also chose the

Bar Chart to present the distribution of customers by

region in terms of total number. This visualization is

particularly useful for decision-makers because it

makes it easier to identify regions with a lower

number of customers (Airinei & Homocianu, 2010).

The analysis of the bar graph makes it possible to

direct marketing efforts towards less active regions,

targeting these potential customers with strategies

adapted to social networks.

Choosing Visualisation for PM:

The first example explores the social network (Turner

et al., 2012) of employees. This analysis allows to

visualize the interactions between employees, to

identify the poles of influence and to understand the

collaborative dynamics within the organization.

The second example implements a Process Map

(Turner et al., 2012), to analyse the sales management

process, as shown in figure 6. This approach makes it

possible to understand the progress of sales steps,

detect inefficiencies and implement optimization

strategies to improve sales performance.

ICSOFT 2024 - 19th International Conference on Software Technologies

454

Figure 6: Sales management handling process map.

6.1.3 Choosing How to Represent

Dashboards

Once individual visualizations are developed, the

next step is to integrate them into dashboards in a

clear, concise and visually appealing way. The goal is

to allow users to quickly and easily understand the

data presented. As part of this project, Power BI tool

was chosen to create those dashboards. We have the

most relevant KPIs, which are quantifiable and

measurable, for the goals of our dashboards. Color

and font are also used to highlight important

information and create a visual hierarchy by ensuring

consistency with the corporate graphic charter. In

addition, we keep the layout simple and clean by

trying not to clutter the dashboards with too much

information by focusing on the most important

elements and leaving enough white space for easy

reading. Figures 7, 8 and 9 illustrate the intermediate

DM, BI and PM dashboards obtained, respectively.

Figure 7: Obtained intermediate DM dashboard.

Figure 8: Obtained intermediate BI dashboard.

Figure 9: Obtained intermediate PM dashboard.

6.2 Global Dashboard

The analyst uses the results of the guidance tool to

develop the overall dashboard, which is structured in

three distinct sections, devoted respectively to DM,

PM and BI. Each section presents a defined number

of visualizations, carefully selected to ensure optimal

readability.

In our specific case study, the BI cards were

selected because of their ability to present clear and

useful values. The PM sales management process

map was chosen to illustrate the main sales process,

while the social network allows employees to see

their activities. Finally, the table describing the

SimpleKMeans DM algorithm is chosen for its ability

to confirm decision tree results and provide additional

A Decision-Making Approach Combining Process Mining, Data Mining and Business Intelligence

455

Figure 10: Obtained global dashboard.

information. The global dashboard, which brings

together visualizations from different areas of

analysis, provides a comprehensive overview of the

company’s performance and the factors that influence

it. This holistic approach allows the manager to make

informed decisions based on evidence and a thorough

understanding of different aspects of the business (Fig

10).

The expert then has the freedom to customize the

dashboard according to his/her needs. For example, if

he/she wants to replace a table with a decision tree

visualization, Power BI gives the flexibility to make

this change with ease: simply delete the dashboard,

navigate to the DM Intermediate Dashboard and

transfer the desired visualization to the global

dashboard.

In summary, a global dashboard that combines

visualizations from data mining, business intelligence

and process mining is a powerful tool for companies

looking to improve their performance, make informed

decisions and optimize their operations.

6.3 Validation

A thorough discussion with the expert highlighted the

many benefits of the proposed approach. The expert

first welcomed the notion of intermediate dashboards,

highlighting their ability to facilitate data exploration

and analysis, avoiding direct immersion in a large

database. Besides the usefulness and

complementarity of each of the intermediary

dashboards (respectively specific to Process Mining,

Data Mining and Business Intelligence), their

combination in the global dashboard seemed

particularly relevant to him. In fact, it offers him/her

a global and coherent vision informing him/her about

the health of the company and the functioning of its

activities. The expert particularly appreciated the

ability to customize the global dashboard by selecting

the visualizations that best suit his needs and

preferences. This flexibility allows focusing on the

most important information and saving time in the

decision-making process.

7 CONCLUSION

A single dashboard that integrates Process Mining

(PM), Data Mining (DM), and Business Intelligence

(BI) visualizations offers a powerful solution to this

challenge. This combined approach empowers

decision-makers to identify trends, anomalies, and

opportunities with greater clarity. Beyond effective

problem-solving, the dashboard facilitates proactive

management by aiding in problem identification,

targeted solutions, and KPI tracking. Ultimately, the

improved decision-making facilitated by this

comprehensive data analysis leads to enhanced

overall company performance. This paper outlines

our work, which adheres to this data-driven approach

and proposes the creation of a robust global

dashboard harnessing the strengths of PM, DM, and

BI. We first create intermediate dashboards which

offer independent data structuring and analysis within

each of the domains of PM, DM and BI. Second, we

develop a Powerful Global Dashboard which is a

central hub integrating visualizations from all three

domains, delivering a coherent and interactive

overview. This comprehensive approach fosters a

ICSOFT 2024 - 19th International Conference on Software Technologies

456

data-driven decision-making ecosystem, allowing for

informed choices that propel business forward.

As part of our future work, we propose to extend

the scope of our project by integrating case studies

from other sectors of activity, such as e-health. This

challenge represents an essential strategic step to

enrich our project, broaden its scope of application,

strengthen its credibility and contribute to the

advancement of knowledge in the field of data

analysis.

In order to further optimize the user experience

and decision-making, we consider the integration of

artificial intelligence (AI) as a major development

axis. AI will enrich dashboards and provide valuable

decision support to users. Specifically, we will

explore the use of machine learning techniques such

as time series analysis algorithms to identify trends

and anomalies in real time. In addition, natural

language processing (NLP) could be integrated to

analyze the texts present in the data and thus enrich

the insights extracted. With these AI technologies,

users will receive immediate feedback and alerts

based on real-time analysis of dashboard content.

REFERENCES

Airinei, D., & Homocianu, D. (2010). Data visualization in

business intelligence. Proc. of the 11th WSEAS Int.

Conf. on Mathematics and Computers in Business and

Economics, MCBE ’10, Proc. of the 11th WSEAS Int.

Conf. on Mathematics and Computers in Biology and

Chemistry, MCBC ’10, 164–167. https://doi.org/

10.4324/9781315471136-6

Azzam, T., Evergreen, S., Germuth, A. A., & Kistler, S. J.

(2013). Data visualization and evaluation. New

Directions for Evaluation, 2013(139), 7–32.

https://doi.org/10.1002/ev.20065

Breşfelean, V. P. (2007). Analysis and predictions on

students’ behavior using decision trees in weka

environment. Proceedings of the International

Conference on Information Technology Interfaces, ITI,

51–56. https://doi.org/10.1109/ITI.2007.4283743

Hendricks, R. (2019). Process Mining of Incoming Patients

with Sepsis. Online Journal of Public Health

Informatics, 11(2). https://doi.org/10.5210/ojphi.v11

i2.10151

Kumar SM, & Meena Belwal. (2017). Performance

Dashboard Cutting-Edge Business Intelligence and

Data Visualization. IEEE.

Martinez-Millana, A., Lizondo, A., Gatta, R., Vera, S.,

Salcedo, V. T., & Fernandez-Llatas, C. (2019). Process

mining dashboard in operating rooms: Analysis of staff

expectations with analytic hierarchy process.

International Journal of Environmental Research and

Public Health, 16(2). https://doi.org/10.33

90/ijerph16020199

Maya D. Albayrak, & William Gray-Roncal. (2019). Data

mining and sentiment analysis of real-time Twitter

messages for monitoring and predicting events.

Nik, M. R. H., Van Der Aalst, W. M. P., & Sani, M. F.

(2019). BIPM: Combining Bi and process mining.

DATA 2019 - Proceedings of the 8th International

Conference on Data Science, Technology and

Applications, 123–128. https://doi.org/10.5220/0

007741901230128

Orlovskyi, D., & Kopp, A. (2020). A Business Intelligence

Dashboard Design Approach to Improve Data

Analytics and Decision Making.

Schuh, G. , Reinhart, G. , Prote, J. P. , Sauermann, F. ,

Horsthofer, J. , Oppolzer, F. , & Knoll, D. (2019). Data

mining definitions and applications for the

management of production complexity.

Soma Ajibade, S., & Adediran, A. (2016). An Overview of

Big Data Visualization Techniques in Data Mining. In

International Journal of Computer Science and

Information Technology Research (Vol. 4).

www.researchpublish.com

Tripathi, A., Bagga, T., & Aggarwal, R. K. (2020).

Strategic Impact of Business Intelligence: A Review of

Literature Abstract.

Turner, C. J. , Tiwari A, & Olaiya R. (2012). Business

Process Mining: From Theory to Practice.

Van der Aalst, W. (2016). Process mining: Data science in

action. In Process Mining: Data Science in Action.

Springer Berlin Heidelberg. https://doi.org/10.100

7/978-3-662-49851-4

A Decision-Making Approach Combining Process Mining, Data Mining and Business Intelligence

457