Journal of Artificial Intelligence and Capsule Networks (2020)

Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

Data Mining Based Marketing Decision

Support System Using Hybrid Machine
Learning Algorithm
Dr. T. Senthil Kumar,
Associate Professor,
Computer Science and Engineering Department,
Amrita School of Engineering,
Coimbatore, TamilNadu, India.
Email: [email protected]

Abstract: - Data mining is widely used in engineering and science, On the contrary, it is used in finance and marketing
applications to resolve the challenges in the respective fields. Data mining based decision support system enhances the
organization performance by analysing the ground reality. Turbulent economy is common for every organization due to the
competition, cost, tax pressures, etc., Privatization, Globalization and liberalization drags the organization more into a
competitive environment. In order to balance the competition and withstand to achieve desired gain proper marketing strategies
are need to planned and executed. Marketing decision support system helps to reduce the organization burdens in analysing
and strategical planning through its efficient data mining approach. This research work proposed a data mining based decision
support system using decision tree and artificial neural network as a hybrid approach to estimate the marketing strategies for
an organization.

Keywords: Data Mining, Marketing decision support system, Decision Tree, Artificial Neural Network.

1. Introduction

Due to the technology development, computer based business decision is commonly adopted even in small
scale organizations to obtain a clear picture about the organization in next few years. Based on the analysis, the
market value of an organization increases which provides more benefit to the owners and stack holders. Decision
support systems are evolved to reduce the human burden in analysing the organizational data. It provides quick
decisions based on the available information which could help to improve the organization quality with in short
duration. Since, data quality defines the quality of an organization, so that decision could be taken at right time
which diverts the growth towards right path. Decision support systems can able to select the data and analyse to
obtain the right trends, so that strategies and solutions could be framed. Data is the main element in decision
process so that the support system summarizes the information either graphically or textually using its expert
artificial intelligence. Decision support systems collects the typical information such as assets, legacy and
relational data and prepares comparative analysis as assumptions of future trend in a specific context. Most of the
organizations adopted this decision support system as a day to day activity.

The common practice followed by an organization to collect the sales data at regular interval of time along
with budget and forecast details in order to update the marketing strategy. Similarly, decision support systems
collect information, analyse and reorganize the data to obtain constructive strategical decisions. Figure 1 depicts
the general illustration of decision support system.

185
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020
Journal of Artificial Intelligence and Capsule Networks (2020)
Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

Fig. 1 Decision support system

Generally, decision support systems are differentiated into active and passive models. In which passive
models collects the information and organize them in an effective manner and it do not provide any suggestions
or decisions from the collected information. In case of active decision support system, it collects and process the
information to produce solutions and strategies. On the other, cooperative decision support collects and analyse
the data and along with human interaction it refines the solutions to obtain best strategies. Based upon the
applications, decision support system is categorized into

 Data Driven Decision support system,

 Communication driven decision support system,
 Document driven decision support system,
 Model driven decision support system
 Knowledge driven decision support system.

Data driven decision support system concentrates on data collection process initially, since collecting
essential information helps to obtain better decision with in short duration. Once the data is collected, it is
manipulated based on the user needs, since the data may be of structured or unstructured due to internal and
external sources it is essential to manipulate the data. A time series is framed to manipulate the internal data and
if needed for external data also. Data driven decision support system provides simple file access system which
could be accessed and retrieved as queries. Managers, employees, product suppliers and service persons are the
target persons in data driven model to collect essential information. In case of communication driven decision
support system analyses the shared tasks where more than a person is working on a single process. in order to
obtain better strategy different collaborator are allocated to work together in communication driven models.
Internal teams, partners are the main targets of communication driven model. Web based or client server based
technology is widely adapted as communication driven process which provides decision support.

Document driven decision support systems performs operation based on group of users. Searching web pages
and identifying relevant documents for a given specific set is the process in document driven models. Retrieval
and analysis of documents is depending on storage and technology adaptation in this models. Variety of document
types such as text document, database records and spreadsheets, etc., are processed to obtain decision in document
driven models. Similar to communication driven, document driven decision support system also uses web based
or client server based technology to obtain decision. Model driven decision support systems has option to

186
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020
Journal of Artificial Intelligence and Capsule Networks (2020)
Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

manipulate the financial data to obtain optimized solution for organization growth. Generally, model driven
decision support system does not require large dataset since it performs quantitative analysis and provides
elementary functionalities using limited parameters. People interact with an organization, managers, staffs are the
main targets of model driven decision support systems which could be executed in a client server or standalone
personal computers.

Knowledge driven decision support systems has wide range of user coverage in an organization which
includes others also to interact. Management advice is an essential process in knowledge driven models so that
the organization could select the right service or products. It can also suggest actions to the managers. Knowledge
driven model are like personal computers with problem solving skills so that understanding the problem in that
domain could be identified easily. Benefits of decision support system is depicted in figure 2 along with its
limitations.

Merits Demerits

 Improved decision reliability.  Limited to computer system

 Ability to process and design.
information and knowledge.  Not efficient as human
Decision decision making includes
 Minimum time to obtain
decision. Support creativity and imagination.
 Solution for largescale, System  Natural language processing
complex issues. is not sophisticated.
 Strategical advantage over  Normally designed for
other organizations. narrow applications,
 Evidence generation to multiple decision making is
support the decision. quite difficult.

Fig. 2 Merits and Demerits of Decision support system

In other ways decision support systems employed in agriculture, where the farmers use this decision support
for crop planning. Based on analysis, it provides suitable cultivating time, fertilize and harvest time which helps
the farmers to plan accordingly. In case of medical field, decision support system is used as clinical decision
support system which is used to manage the complex clinical data. Since medical data has wide range of versatile
information such as tracking orders, preventive protocols, these decision models are employed as cost control
models which reduces the repetitive, duplicate tests and drug purchase. Decision support system is used to predict
the natural disasters like floods based on analysing the real time weather data and historical data. So that possible
flood regions could be identified which avoid huge property loss and human causalities. Decision support system
in educational institutions helps to obtain details about student details. Based on prediction it could define the
students count for the upcoming year. Depends upon the results, institutions plan and process the student
enrolment process to obtain full strength to run a course.

2. Related works

A vast survey has been made to obtain prior knowledge on issues in decision support system. based on
application domain, system design and complexity, existing models are analysed and presented in this section.
Daniel.et.al. [1] discussed about smart maintenance decision support system to manage corporate data. The
research work brought insights on big data analytics and its decision making process. using a cost minimization
algorithm, the proposed work discussed the issues in industry large scale data. Research work attains better
predictive results for the case studies using its analytical decision process. In order to reduce faults in the
production systems, knowledge and experience are the key factors. Sebastian.et.al. [2] considered those key

187
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020
Journal of Artificial Intelligence and Capsule Networks (2020)
Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

factors in research work to obtain decision for distributed production environment. Since knowledge sharing is an
important process for test and service teams which is directly related to quality of an organization, research work
provides better decision support system using test cases. Incorporating test cases with fault symptoms the
maintainers could able to identify the faults. İhsan Erozan.et.al. [3] proposed a decision support system for
manufacturing industries to manage the activities. Parameters such as utilization rate, duty cycle, redundancy,
stock effects are considered to define the proposed work which improves the maintenance performance and
product reliability. In order to maintain critical components activities proposed decision support system is tested
with industrial datasets which obtains better reliability compared to conventional decision models.

Francesco.et.al. [4] reported the issues in group decision support systems through k-means algorithm.
Combined with hierarchical process research work attains remarkable decisions for the given tasks. In order to
obtain the decision, alternatives are sorted and ordered based on its importance and Bezier curve fitting is used to
obtain the decision makers preference function. The knapsack model used in this research work chooses
alternatives based on the classes and budget constrains which yields better performance. Similarly, web based
group decision support system is reported in Siddiqui.et.al. [5] research model. Research work is carried out to
obtain solution for team preparation issue in a business school. Considering the course offerings, stakeholder
information, assignments, timetable preparation, the observations are projected as multi stage issues. Proposed
model used web application to define the workflow and automates the timetabling task using optimization model.
Conventional spread sheet based tools could be replaced with improved quality and efficient workflow are the
advantages of this decision support system.

Morteza.et.al. [6] considered the stability, delivery status, service attempts and financial growth to define the
decision making system. To obtain the similarity and deployment of quality function these decision making is
developed in the research work. Agricultural partners and investors are the target factors of this group decision
making process. proposed research model provides an integrated and efficient decision making system which
improves the reliability and quality while obtaining decisions. Katia.et.al. [7] reported the issues in agricultural
production through decision support system as industrial processing based on the environmental, social and
economic dimensions. Proposed model used evaluation mechanism to define the efficiency, cost and production
factors. Sustainability index and semaphore are used to maintain the sustainability in the proposed approach.

Irrigation decision making and its related issues are reported in Nicholas.et.al. [8] research work. Heed
decision theory and decision modelling is proposed in the research work to distinguish the decision support system
tasks and real time decisions. In order to process and obtain business solutions decision modelling notation is
suggested, decision ontology is suggested for web modelling and decision modelling ontology is suggested for
conventional ontology. Yang Luo.et.al. [9] reported the issues in data centre and its decision support system to
evaluate the energy and temporal availability. Proposed work is framed to analyse the issues in medium sized
commercial data centre using four stage framework. Considering the cost benefits, environmental impact.
Proposed decision support system is designed to deliver an optimized heat recovery process. Through proposed
approach the overall energy requirement of data centre is streamlined.

Rule based reasoning is proposed in Mahsa.et.al. [10] research work along with fuzzy logic classifier to obtain
the triage level of patients. These clinical decision support system provides prediction results based on emergency
service index which has patient vital signs. These vital signs are considered as inputs to fuzzy logic to predict the
triage levels. Similar clinical decision support system is reported in Ahmed.et.al. [11] research work which uses
l-means clustering for decision making process. Processing the leukemia cells, the features are extracted to obtain
decision based on the internal structure. Proposed work is experimentally verified using public dataset which
effectively identifies the leukemia cells compared to conventional clinical decision support systems. Another
clinical decision support system is reported by Marieke.et.al. [12] to measure the plasma concentration in therapy.
By measuring the plasma concentration, patient health condition is efficiently reviewed using proposed decision
support system.

Muhammad.et.al. [13] reported the issues in rubber production through decision support system to identify
the yields in a smallholder rubber production unit. Research model considers the institutional factors and social
impacts on rubber production through decision support system which improves the profitability and viability of
the farm management. Zouhair.et.al. [14] proposed a proactive decision support system to predict the crash events
on vehicle kinematics. Considering the roadway geometrics, driver inputs and real time weather data proposed

188
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020
Journal of Artificial Intelligence and Capsule Networks (2020)
Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

decision support system is developed through machine learning approach. High prediction performance is
achieved through the proposed approach compared to data balancing techniques.

Bon-Gang Hwang.et.al. [15] proposed a knowledge based decision support system to identify the decision
factors for prefabricated prefinished volumetric construction. Based on the access feasibility and adoption factors
these decision support system is developed to attain better decisions to implement prefabricated prefinished
volumetric construction in Singapore. From the above survey it is observed most of the decision support system
are processed through conventional statistical models. Very less research works are processes through machine
learning models and the efficiency of the system could be improved further. Considering the above issues, the
proposed model is designed as a hybrid approach which efficiently provides decision support to the marketing
domain.

3. Proposed work

The proposed hybrid model is developed as a combination of decision tree and artificial neural network
model. This section provides the necessary mathematical formulation for the proposed hybrid algorithm. The
proposed Hybrid model is applied in a manufacturing company to provide a decision towards their marketing
ideas considering the issues present in the company and its marketing section. The choice of retained company
which leads into a poor image in the environment and stack holders, so it is essential to obtain a decision support
system to development the production and marketing strategies, so that the company profit could increase. In
order to assess the actual situation different pressures are considered which increases the strategical benefits. A
list consist of all the necessary criteria is designed and sent to the expert team and marketing team. Reponses from
the experts and marketing team are analysed further to obtain the necessary sub-criteria for the proposed system.
Based on the analysis again separate list is generated and it is communicated to everyone in the team to obtain the
feedback over the selected criteria. In the last, the consensus is selected based on the sub-criteria which is used to
refine the industry production and marketing process. process flow of the proposed hybrid model is depicted in
figure 3, which has three phased of operation as problem setting, model development, action developments.

Fig.3 Proposed Hybrid Model

The initial problem formulation phase is used to define the scope and identifying the unavoidable issues in
the industry. The route-cause analysis and its relationship is essential for decision support system, in order to
identify the root a framework is setup for production and marketing as objectives. In this level, the essential

189
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020
Journal of Artificial Intelligence and Capsule Networks (2020)
Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

production and marketing objectives are further subdivided in order to reach the framework from top level to
bottom level employees. Based on hierarchical order, these objective functions are created.
Model development is the second phase which includes formal models. These models are developed based
on the sub-objectives, so that the present manufacturing and marketing aspects could be rearranged. Considering
the normalizing, constituting and decision tree weight functions which is given as a matrix function

𝐷 = [𝐷𝑖𝑗 ]𝑛×𝑚 (1)

where the column of the matrix is used to represent the alternate solutions and row of a matrix is used to
represent the triggered pressures. The final expanded matrix function for the given objective is
𝐷11. … . 𝐷1𝑗 … . 𝐷1𝑚 .
.. .. ..
. .. .
| |
𝐷 = 𝐷𝑖1.. … . 𝐷𝑖𝑗 … .
..
𝐷𝑖𝑚
.. (2)
| ... .. |
.
..
𝐷𝑛1 … . 𝐷𝑛𝑗 … . 𝐷𝑛𝑚

The Normalized decision matrix is given as

𝐷𝑁 = [𝐷′𝑖𝑗 ]𝑛×𝑚 (3)

𝐷𝑖𝑗
where 𝐷′𝑖𝑗 = , 𝑖 = 1,2,3 … . 𝑛, 𝑗 = 1,2,3 … . . 𝑚.
√∑𝑛
𝑘=1 𝐷′𝑘𝑗

Artificial neural network is used to refine the results of decision tree model. Figure 4 depicts the neural
network model used in the proposed decision support system for classifying the decision tree hierarchical results.

Fig.4 Neural Network Model

The process starts from the three level inputs from the decision tree system and then it is processed into
9 different intermediate level based on the sub-objectives defined for the proposed decision support system. The
interlinked process indicates the proposed system has interaction with top level to bottom level of employee in an
organization. Based on the sub-objective it is cumulated and classified to obtain the final strategical plan for the
manufacturing and marketing process.

190
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020
Journal of Artificial Intelligence and Capsule Networks (2020)
Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

4. Result and Discussion

The proposed hybrid model is analysed based on the criteria such as internal pressure, criteria, targets and
external pressures. So that each category is evaluated against the proposed algorithm and compared with
conventional hidden Markov and support vector machine based decision support systems. Main objectives are
given as A1 for internal pressure, B1 for external pressures and C1 for target and production criteria, where the
sub-objectives are given as SC1 for job satisfaction (Common for production and marketing employees), SC2 for
cognitive skills, SC3 for employee contribution to organization, SC4 for employee power, SC5 for individual
efforts, SC6 for timely response, SC7 for available resource, SC8 for organization culture, SC9 for maturity towards
company activities. The classification results are clustered into C1, C2, C3 and then final classification results are
obtained as output of decision support system for organization growth.

Table 1. Decision Tree Matrix

SC1 SC2 SC3 SC4 SC5 SC6 SC7 SC8 SC9
A1 2 2 2 4 4 4 8 5 5
B1 5 5 4 5 5 5 6 3 7
C1 6 8 7 6 5 6 7 8 6

Table 1 provides details about decision tree matrix values for the clusters with respect to the main objectives
and sub-objectives. It is observed that the external pressure is almost same for all the three categories. On contrary,
the internal pressure is much greater for sub-objectives. In terms of individual efforts, the attributes are high
compared to others which depicts that the employees are willing to contribute their skills to improve the
organization as a team and an individual.

Table 2 Clustering Comparison for the Criteria

C1 C2 C3

Priority
SC1 SC2 SC3 SC4 SC5 SC6 SC7 SC8 SC9 Function
values

SC1 1 1 2 3 4 2 4 1 1 0.1654

C1 SC2 1 1 2 3 4 2 3 1 1 0.1654

SC3 1 2 3 1 4 1 2 1 2 0.1452

SC4 1/2 1/2 1 3 3 2 1 3 3 0.0564

C2 SC5 1/3 1/2 1/3 1 2 2 1/4 1/4 1/2 0.0562

SC6 1/4 1/4 1/3 1 2 2 1/2 1/2 1/4 0.0456

SC7 1/3 1/3 1/3 2 3 3 1/4 1 1/4 0.0445

C3 SC8 1/2 1/2 1 4 2 1 1 1/4 4 0.0812

SC9 1/2 1 1/2 4 2 1 1/4 1/2 1 0.1256

Table 2 depicts the clustering comparison details for the criteria and it is built based on the response of
objectives and sub-objectives from managers to low level employees. The priority function value is obtained based
on the consistency of the clustering and threshold values. It is acceptable if the value is below 0.1 and others are
considered as non-acceptable.

191
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020
Journal of Artificial Intelligence and Capsule Networks (2020)
Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

Fig. 5 Priority assessment comparison

Based on the weights the priority values of the proposed system are measured and compared with hidden
Markov model and support vector model and it is depicted in figure 5. It is observed that proposed model has
better priority values by assessing the employees, stack holders and management opinions. While the other models
have less priority values due to its inefficient assessment strategies.

5. Conclusion

A hybrid decision support system is proposed in this research work using decision tree and artificial neural
network. The issues in existing decision support models are identified to develop this decision support system to
improve the manufacturing and marketing strategies of an organization. Based on the option obtained from the
employees, the main objective function is subdivided into sub-objective function to analyse the organization from
top level to bottom level. Given responses are analysed using decision tree matrix and trained using artificial
neural network to obtain better classification strategies. Experimental results are observed and compared with
conventional hidden Markov Model and support vector based decision support system. Proposed model attains
better classification results compared to other model which could be applicable to investigate various domains.
Further this research work could be improved by including an optimization models to improve the assessment
performance.

References

1. Daniel Bumblauskas, Douglas Gemmill, Amy Igou, Johanna Anzengruber (2017). Smart Maintenance
Decision Support Systems (SMDSS) based on corporate big data analytics. Expert Systems with
Applications.90: 303-317.
2. Sebastian Abele, Michael Weyrich (2017). Decision Support for Joint Test and Diagnosis of Production
Systems based on a Concept of Shared Knowledge. IFAC-PapersOnLine. 50(1): 15227-15232.
3. İhsan Erozan (2019). A fuzzy decision support system for managing maintenance activities of critical
components in manufacturing systems. Journal of Manufacturing Systems. 52:110-120.
4. Francesco Lolli, Alessio Ishizaka, Rita Gamberini, Bianca Rimini, Laura Prandini (2017). Requalifying
public buildings and utilities using a group decision support system. Journal of Cleaner Production.164:
1081-1092.

192
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020
Journal of Artificial Intelligence and Capsule Networks (2020)
Vol.02/ No. 03
Pages: 185-193
http://irojournals.com/aicn/
DOI: https://doi.org/10.36548/jaicn.2020.3.006

5. Atiq W. Siddiqui, Syed Arshad Raza, Zeeshan Muhammad Tariq (2018). A web-based group decision
support system for academic term preparation. Decision Support Systems. 114:1-17.
6. Morteza Yazdani, Pascale Zarate, Adama Coulibaly, Edmundas Kazimieras Zavadskas (2017). A group
decision making support system in logistics and supply chain management. Expert Systems with
Applications.88: 376-392.
7. Katia Regina Evaristo de Jesus, Sérgio Alves Torquato, Pedro Gerber Machado, Catiana Regina Brumatti
Zorzo, Dilvan A. Moreira (2019). Sustainability assessment of sugarcane production systems: SustenAgro
Decision Support System. Environmental Development. 32:1-15.
8. Nicholas J. Car (2018). USING decision models to enable better irrigation Decision Support Systems.
Computers and Electronics in Agriculture.152:290-301.
9. Yang Luo, John Andresen, Henry Clarke, Matthew Rajendra, Mercedes Maroto-Valer (2019). A decision
support system for waste heat recovery and energy efficiency improvement in data centres. Applied Energy,
250: 1217-1224.
10. Mahsa Dehghani Soufi, Taha Samad-Soltani, Samad Shams Vahdati, Peyman Rezaei-Hachesu (2018).
Decision support system for triage management: A hybrid approach using rule-based reasoning and fuzzy
logic. International Journal of Medical Informatics. 114:35-44.
11. Ahmed S. Negm, Osama A. Hassan, Ahmed H. Kandil (2018). A decision support system for Acute
Leukaemia classification based on digital microscopic images. Alexandria Engineering Journal.57(4):
2319-2332.
12. Marieke Pereboom, Inge J. Mulder, Sjoerd L. Verweij, Ruud T. M. van der Hoeven, Matthijs L. Becker
(2019). A clinical decision support system to improve adequate dosing of gentamicin and vancomycin.
International Journal of Medical Informatics. 124:1-5.
13. Muhammad Fadzli Ali, Ammar Abdul Aziz, Siti Hawa Sulong (2020). The role of decision support systems
in smallholder rubber production: Applications, limitations and future directions. Computers and
Electronics in Agriculture. 173:1-14.
14. Zouhair Elamrani Abou Elassad, Hajar Mousannif, Hassan Al Moatassime (2020). A proactive decision
support system for predicting traffic crash events: A critical analysis of imbalanced class distribution.
Knowledge-Based Systems. 205:1-14.
15. Bon-Gang Hwang, Ming Shan, Kit-Ying Looi (2018). Knowledge-based decision support system for
prefabricated prefinished volumetric construction. Automation in Construction. 94:168-178.

193
ISSN: 2582-2012 (online)
Submitted: 29.07.2020
Accepted: 20.08.2020
Published: 28.08.2020