Integration of Artificial Intelligence

Technology in Management Accounting

Information System: An Empirical Study

Emon Kalyan Chowdhury

Abstract At present, most of the business organizations take their management

decisions using traditional approach. In the traditional approach, the freedom to be
flexible is limited due to numerous assumptions. This paper aims to establish an
artificial neural network-based model to predict management information and verify
the accuracy of the model using some real data. The proposed model covers five
dimensions, namely, accounting analysis management system, accounting decision
support system, performance management information system, risk management
information system, and environmental management information system. It is
observed that the proposed model can predict the management accounting informa-
tion by 98.83%, which is extremely good and meets the accounting information
requirement.

Keywords Artificial intelligence · Machine learning · Management accounting ·

Information system · Neural network

1 Introduction

Management accounting provides information to managers who make important

decisions in an organization (Garrison et al., 2003). The size and complexity of data
is increasing day by day as a result managers are in serious trouble in processing
large amount of data (Munim et al., 2020). The success of a decision depends on the
quality of the information. Therefore, an efficient management accounting informa-
tion system where data are processed through artificial intelligence technology plays
a vital role in improving the operating efficiency of an organization (Zhang, 2021).
Management enterprises are substantially dependent on advanced information
technology to make rational and effective decisions. Among management informa-
tion systems, the management accounting information system is the most important

E. K. Chowdhury (✉)
CIU Business School, Chittagong Independent University, Chattogram, Bangladesh

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 35

M. Z. Abedin, P. Hajek (eds.), Novel Financial Applications of Machine Learning
and Deep Learning, International Series in Operations Research & Management
Science 336, https://doi.org/10.1007/978-3-031-18552-6_3
36 E. K. Chowdhury

segment (Hutahayan, 2020). The significance of management accounting informa-

tion system lies in the economic progress, expansion, scale of economies, acquisi-
tion, and continuous improvement of strengths through scientific management
decisions (Cai et al., 2019).
Practically, the use of management accounting information system is confined to
the cost management, preparation of different budgets, and performance manage-
ment. Smooth functioning of enterprise management is highly dependent on the
comprehensive and stable construction of management accounting information
systems integrated with other management information systems (Goetz et al., 2015).
The remaining part continues as follows. Section 2 reviews previous studies.
Section 3 analyzes different models based on artificial intelligence technology.
Section 4 experiments the success rates of prediction capacity of model using
authentic management information data, and Sect. 5 concludes the paper.

2 Literature Review

Management control systems ensure optimal use of limited resources to achieve the
organization’s goal. In addition to financial data, an efficient management control
system also uses psychological and control variables (Nguyen et al., 2017). The data
from multiple sources are collected and fed into the management information system
so as to generate various sub-objectives from a single organizational objective. It
helps to compare the actual performance with the projected plans from diverse
perspectives (Al-Ali et al., 2017). To sustain itself in a competitive and
technology-based environment, an organization must strengthen its managerial and
supervisory functions by introducing a management control system (Chi et al., 2019;
Xin et al., 2018). Out of the different wings of the management information system,
the development of the management accounting information system is crucial, as it
directly contributes to the organization’s financial solvency, internal control system,
customer retention, and overall sustainability (Chowdhury, 2019; Ward et al., 2016).
Recently, the use of an e-commerce-based accounting information system has
increased tremendously among the enterprises to enjoy competitive advantages
(Shajalal et al., 2021; Hidayat et al., 2020). Management accounting plays an
important role in fulfilling the economic needs of an organization’s operation and
management with the help of responsibility center. The responsibility center ensures
optimum uses of internal accounting control systems and further assists in organiz-
ing and delivering other functional internal management systems (Ghasemi et al.,
2019). Amershi et al. (2014) observed a significantly positive impact of management
accounting on innovation management. Management accounting systems simplify
the cost calculation of single and batch products (Rodriguez-Galiano et al., 2015).
Cooper et al. (2017) noticed the increasing popularity of using balanced scorecards
in organizations to measure the performance of different indicators.
The traditional management accounting system mostly depends on the assump-
tions rather than versatility of data, which imperatively directs to take fixed
Integration of Artificial Intelligence Technology in Management. . . 37

decisions. This study finds a gap to explore the possibility of taking dynamic
decisions by using alternative models where artificial intelligence technology is
used in line with machine learning and data mining algorithms.

3 Artificial Neural Network (ANN)

The design of ANN is inspired by the structure of biological neurons such as the
human brain. In a human brain, neurons create a network through interconnections.
A neuron is known as a cell and executes a single task by responding to an input
signal. In an ANN, the nodes are connected to each other and establish a network
among themselves. The nodes are designed using artificial intelligence to handle
massive amount of data using multiple equations simultaneously. In this network,
the equations are established through sequential computations following a trial-and-
error approach (Abedin et al., 2021; Chakraborty et al., 2018). The basic structure of
ANN is expressed in Fig. 1.
Input neurons X1, X2, . . ., Xn indicate various inputs to the network, synapse
weights W1, W2, . . ., Wn signify the weights of connections. The weights are very
important in ANN as these represent the strength of each node. The weights that
govern the effect of neurons are measured in the numerical parameters, which
determines the output by converting the input.
The hidden layer performs the processing task. It applies two operational func-
tions, the summation function and the transfer or activation function. The summation
function multiplies each input (Xi) with the corresponding weight (Wi) and all
products (Wi × Xi) result in the summation function ξ = ∑Wi × Xi + B, where
B represents the bias value. It controls the output of the neuron in line with the
weighted sum of inputs.
The activation function transforms the input signal from the summation function
into to output of a node for an ANN model. Each ANN is made up of three
components. First, the node character determines inputs and outputs through signal
processing. Second, the network topology determines how the nodes are connected

X1 Bias
W1
X2 W2 summation
function
… ξ=∑Wi Xi
Y
transfer/activa-
tion function
Xn Wn Y=∫(ξ)

Fig. 1 Model of an artificial neuron

38 E. K. Chowdhury

Legend Increase
revenue
+
+ Decrease
Goal costs
Increase
Sales

-
+ Yearly
Refinement Yearly

+
Strong
+
Acquire costs
sales
economic other
AND OR growth
Stay +
competitive companies
Influence

+
Provide

+
+P State-of-the-
Have a Range of Weaker
art transaction

+
+ worldwide Services US Dollar
systems
Task Handle presence
+

-
transaction Offer Offer
+P

volumes cards internation Minimize

Indicator al banking P international
+

Accurate Select conversion

transactions Facilitate type of costs
Evaluates
card card(s)
# Data processing Offer jointly
+

Measures International
entry Offer Offer branded
errors charge credit international conversion
Situation Credit card cards cards currency cards costs
(External) International transaction
Development
Make Translate
Situation International Collect
Collect agreements revenue and
(Internal) development Subscription with other credit
Interest costs across
program fee card companies currencies

Fig. 2 Business intelligence model for a credit card company. Source: Horkoff et al. (2012)

and organized. Third, the learning rules create and adjust weights. A few widely used
ANN-based models have been discussed below.
Business Intelligence (BI) Model
BI helps an organization excel at operational activities in such a way that helps tap
the opportunities in the market while overcoming potential threats. It has the
capacity to generate effective information to take strategic decisions by processing
massive volume of data. BI establishes a network between an organization and the
external environment with the support of different reasoning techniques that controls
influences, situations, and the indicators (Fig. 2). The reasoning tools for this model
are “what if” a bottom-up approach and “is it possible?” a top-down approach
(Horkoff et al., 2012).
Three-Tier Data Model
The three-tier data model is widely used in the data warehouse management of an
organization. It provides subject-wise analytical environment in the global context
(Abedin et al., 2020; Lau et al., 2018). The three tiers have been outlined below:
(a) Conceptual model: This is the top level of the model which is expressed by
topics. Topics are derived from the dimensions and measures. Dimension refers
to a perspective through which people observe the world, and measurement is
related to data information of the dimension. For example, sales volume.
(b) Logical model: Logical model may be classified into two models such as the star
model and the snowflake model. The star model includes the fact table and
dimension model, while at the same time, they are connected to each other. The
Integration of Artificial Intelligence Technology in Management. . . 39

Data dimension
Year Customer ID
Quarter Name
Month Address
Week Sales Fact
Sales ID City
Day
Customer ID
Product ID
Date ID
Store dimension Store ID
Product dimension
ID Sales units
Name
City
Description
State
Price
District
Brand
Zip

Fig. 3 Star model

Data dimension
Year
Customer ID
Quarter
Name
Month
Sales Fact Address
Week
Sales ID City
Day
Customer ID
Product ID
Store dimension Date ID
ID Store ID Product dimension Product variant
City Sales units Name Variant ID
State Description Variant name
District Price Product type
Zip Brand

Fig. 4 Snowflake model

star model is shown in Fig. 3. The sales data are generated in different time
dimension including customer details, store details, and product details.
A snowflake model is an extension of the star model. It includes additional
information about a particular dimension (Fig. 4). It uses similar disk space, is
easy to install, and reduces query performance for multiple tables.
Extract, Transform, Load (ETL) Model
In this model, data are extracted from multiple source systems and then converted to
final data after necessary calculations. The converted data are loaded into the data
warehouse system for managerial decision. Source points include relevant stake-
holders such as analysts, developers, testers, and top brass executives. Since ETL
activities occur regularly, the data warehouse required to be updated, agile, and
properly documented. ETL helps to make critical business decisions, and compare
40 E. K. Chowdhury

Oracle

SQL

Staging Area Data Warehouse

Teradata

Flat file

Fig. 5 ETL model

the data of the source and target system through data migration and manipulation.
Where the transactional database fails to answer complex business questions, ETL
can easily and quickly address them (Hajek & Abedin, 2020; Sabtu et al., 2017).
Figure 5 shows the ETL process in three steps.
In the ETL model, data are fed into the staging area by extracting them from the
source points after due validations. Data are extracted from the source points in raw
format, and at the transformation stage, data are cleaned, mapped, and converted. In
this stage, the ETL assigns values and modifies the data so that business intelligence-
based reports can be generated. Warehousing data is the last step of the ETL model.
Here, a huge volume of data can be loaded in significantly less time. If the loading
process fails, the recovery mechanism is activated without failure of any sort of data
integrity. The entire ETL process is controlled by the warehouse administrator
(Abedin et al., 2018).
Cube Structure
The data cube is a three-dimensional way of presenting data. In this model, the data
are judged from various perspectives. When data cannot be presented in traditional
column and row format due to more variables and context, data cube can make it so
simple by utilizing different angles (Augenstein et al., 2018). Data cubes have the
following categories.
(a) Multidimensional data cube: Most of the online analytical processing (OLAP)
products are designed using a multidimensional array. These OLAPs perform
better than other approaches, as they can be indexed straight to collect subsets of
data. The larger the dimension, the sparser the cubes.
(b) Rational OLAP (ROLAP): This model uses a relational database to store and
manage warehouse data. ROLAP servers are highly scalable and analyze
Integration of Artificial Intelligence Technology in Management. . . 41

massive volumes of data across multiple dimensions. It also stores and analyzes
highly volatile and changeable data.
To understand the presentation of the data in cube structure, the following
information can be considered (Table 1).
The above information is shown in a three-dimensional cube (Fig. 6).
The essence of the cube structure lies in the capacity to show different data in a
single image.
Data Mining (DM) Process
DM is an essential part of the management accounting information system (Kara
et al., 2020). It combines database, statistics, machine learning, and other relevant
technologies. It generates required information for managers amalgamating different
data to enjoy competitive advantages (Abedin et al., 2019). Figure 7 depicts the data
mining process.

4 Proposed Model

In light of the above analysis, this study recommends an Intelligent Management

Accounting Information System (IMAIS) for the decision-making process where the
following aspects are integrated. This model is the extension of Zhang (2021) where
the environmental management information system was not included. In this model,
the impact of the management decision on the environment has been considered. The
integrated systems are as follows:
(a) Accounting analysis management system
(b) Performance management information system
(c) Accounting decision support system
(d) Risk management information system, and
(e) Environmental management information system
This recommended model can provide customized information to take decisions
in time and also helps to run its business in a way better ensuring a sound internal
control system. Figure 8 shows an IMAIS formation structure.
The recommended IAMAIS model covers reporting systems, risk management,
performance management, decision support issues, and environmental issues. Each
sub-system works autonomously and combinedly to fulfill segment and enterprise
requirements.
Test of Model Efficiency
To verify the degree of accuracy of the proposed model, this study has used real
management accounting data. Out of 380 observations, a total of 125 observations
have been used classifying into 13 categories to train the model. The predicted
results and actual results are shown in Fig. 9.
 42

Table 1 Location-wise quarterly data

Location = “Chicago” Location = “New York” Location = “Toronto”
Item Item Item
Home Home Home
Time Ent. Comp. Phone Sec. Ent. Comp. Phone Sec. Ent. Comp. Phone Sec.
Q1 854 882 89 623 1087 968 38 872 818 746 43 591
Q2 943 890 64 698 1130 1024 41 925 894 769 52 682
Q3 1032 924 59 789 1034 1048 45 1002 940 795 58 728
Q4 1129 992 63 870 1142 1091 54 984 978 864 59 784
E. K. Chowdhury
Integration of Artificial Intelligence Technology in Management. . . 43

Fig. 6 Cube structure

Fig. 7 Data mining process

Pre-
Data Validati
processi Analysis
input on
ng

Fig. 8 Intelligent Accounting

management accounting analysis
information system management
system

Environmental Performance
management management
information information
system system

Risk
Accounting
management
decision support
information
system
system

It is observed that the prediction is very close to the actual results for most of the
observations. To get a further clear scenario, the residuals of the actual and predicted
data are shown in Fig. 10.
It is also observed that most residuals hover within 0.05 to -0.05 and a very
insignificant number of observations are above 0.1 to -0.01. This clearly indicates
that the model is capable of predicting management information with an accuracy
rate of 98.83%. As the rate is very close to 100%, it may be applied in the real world.
44 E. K. Chowdhury

1.2

0.8

0.6

0.4

0.2

0
0 50 100 150 200 250 300 350 400

Predicted Actual

Fig. 9 Actual vs. predicted data

0.2

0.15

0.1

0.05

0
0 50 100 150 200 250 300 350 400
-0.05

-0.1

-0.15

Fig. 10 The residuals of actual vs. predicted results

5 Conclusion

This study aimed to formulate a management accounting information system using

machine learning and an artificial neural network model. Being a vital
sub-information system of management information system, the management
accounting information system plays a very important role in the accounting
Integration of Artificial Intelligence Technology in Management. . . 45

development, therefore it should incorporate the accounting analysis management

system, performance management information system, accounting decision support
system, risk management information system, and environmental management
information system. The recommended model can predict the accounting data with
an accuracy rate of 98.83%. As the business world is complex and affected by many
factors, the use of artificial intelligence technology to make management accounting
decisions knows no bounds. It is assumed that the synergy of five dimensions helps
in taking appropriate business decisions. Future researchers may include legal and
ethical issues in the model to make this model more reliable and applicable as these
issues vary from country to country.

References

Abedin, M. Z., Chi, G., Colombage, S., & Moula, F. E. (2018). Credit default prediction using a
support vector machine and a probabilistic neural network. Journal of Credit Risk. Accessed
from https://ssrn.com/abstract=3175776
Abedin, M. Z., Guotai, C., Moula, F. E., Azad, A. S., & Khan, M. S. U. (2019). Topological
applications of multilayer perceptrons and support vector machines in financial decision support
systems. International Journal of Finance & Economics, 24(1), 474–507.
Abedin, M. Z., Chi, G., Uddin, M. M., Satu, M. S., Khan, M. I., & Hajek, P. (2020). Tax default
prediction using feature transformation-based machine learning. IEEE Access, 9, 19864–19881.
Abedin, M. Z., Hassan, M. K., Khan, I., & Julio, I. F. (2021). Feature transformation for corporate
tax default prediction: Application of machine learning approaches. Asia-Pacific Journal of
Operational Research, 2140017.
Al-Ali, A. R., Zualkernan, I. A., Rashid, M., Gupta, R., & AliKarar, M. (2017). A smart home
energy management system using IoT and big data analytics approach. IEEE Transactions on
Consumer Electronics, 63(4), 426–434.
Amershi, S., Cakmak, M., Knox, W. B., & Kulesza, T. (2014). Power to the people: The role of
humans in interactive machine learning. AI Magazine, 35(4), 105–120.
Augenstein, D., Fleig, C., & Maedche, A. (2018, June). Development of a data-driven business
model transformation tool. In International Conference on Design Science Research in Infor-
mation Systems and Technology (pp. 205–217). Springer.
Cai, J., Huang, W., Yang, S., Wang, S., & Luo, J. (2019, August). A selection method for Denoising
auto encoder features using cross entropy. In International Conference on Intelligent Computing
(pp. 479–490). Springer.
Chakraborty, T., Chattopadhyay, S., & Chakraborty, A. K. (2018). A novel hybridization of
classification trees and artificial neural networks for selection of students in a business school.
Opsearch, 55(2), 434–446.
Chi, G., Uddin, M. S., Abedin, M. Z., & Yuan, K. (2019). Hybrid model for credit risk prediction:
An application of neural network approaches. International Journal on Artificial Intelligence
Tools, 28(05), 1950017.
Chowdhury, E. K. (2019). Transformation of business model through blockchain technology.
Accounting and Finance, 47(5), 4–9.
Cooper, D. J., Ezzamel, M., & Qu, S. Q. (2017). Popularizing a management accounting idea: The
case of the balanced scorecard. Contemporary Accounting Research, 34(2), 991–1025.
Garrison, R. H., Noreen, E. W., Brewer, P. C., & Mardini, R. U. (2003). Managerial accounting.
McGraw-Hill/Irwin.
Ghasemi, R., Habibi, H. R., Ghasemlo, M., & Karami, M. (2019). The effectiveness of management
accounting systems: Evidence from financial organizations in Iran. Journal of Accounting in
Emerging Economies, 9(2), 182–207.
46 E. K. Chowdhury

Goetz, J. N., Brenning, A., Petschko, H., & Leopold, P. (2015). Evaluating machine learning and
statistical prediction techniques for landslide susceptibility modeling. Computers and
Geosciences, 81, 1–11. https://doi.org/10.1016/j.cageo.2015.04.007
Hajek, P., & Abedin, M. Z. (2020). A profit function-maximizing inventory backorder prediction
system using big data analytics. IEEE Access, 8, 58982–58994.
Hidayat, A. T., Dewantara, A. M. D., & Saifullah, S. (2020). The development of website on
management information system for e-commerce and services. Jurnal Sisfokom (Sistem
Informasi dan Komputer), 9(3), 380–386.
Horkoff, J., Borgida, A., Mylopoulos, J., Barone, D., Jiang, L., Yu, E., & Amyot, D. (2012,
September). Making data meaningful: The business intelligence model and its formal semantics
in description logics. In OTM Confederated International Conferences. On the move to mean-
ingful Internet systems (pp. 700–717). Springer.
Hutahayan, B. (2020). The mediating role of human capital and management accounting informa-
tion system in the relationship between innovation strategy and internal process performance
and the impact on corporate financial performance. Benchmarking: An International Journal,
27(4), 1289–1318.
Kara, M. E., Fırat, S. Ü. O., & Ghadge, A. (2020). A data mining-based framework for supply chain
risk management. Computers & Industrial Engineering, 139, 105570.
Lau, H. C., Ip, A., Lee, C. K. M., & Ho, G. T. (2018). Development of a three-tier assessment
model: A case study. Benchmarking: An International Journal, 25(7), 2216–2229.
Munim, Z. H., Dushenko, M., Jimenez, V. J., Shakil, M. H., & Imset, M. (2020). Big data and
artificial intelligence in the maritime industry: A bibliometric review and future research
directions. Maritime Policy & Management, 47(5), 577–597.
Nguyen, T. T., Mia, L., Winata, L., & Chong, V. K. (2017). Effect of transformational-leadership
style and management control system on managerial performance. Journal of Business
Research, 70, 202–213.
Rodriguez-Galiano, V., Sanchez-Castillo, M., Chica-Olmo, M., & Chica-Rivas,
M. J. O. G. R. (2015). Machine learning predictive models for mineral prospectivity: An
evaluation of neural networks, random forest, regression trees and support vector machines.
Ore Geology Reviews, 71, 804–818.
Sabtu, A., Azmi, N. F. M., Sjarif, N. N. A., Ismail, S. A., Yusop, O. M., Sarkan, H., & Chuprat,
S. (2017, July). The challenges of extract, transform and loading (ETL) system implementation
for near real-time environment. In 2017 International Conference on Research and Innovation
in Information Systems (ICRIIS) (pp. 1–5). IEEE.
Shajalal, M., Hajek, P., & Abedin, M. Z. (2021). Product backorder prediction using deep neural
network on imbalanced data. International Journal of Production Research, 2021, 1–18.
Ward, L., Agrawal, A., & Choudhary, A. (2016). A general-purpose machine learning framework
for predicting properties of inorganic materials. NPJ Computational Materials, 2, 16028. https://
doi.org/10.1038/npjcompumats.2016.28
Xin, Y., Kong, L., Liu, Z., Chen, Y., Li, Y., Zhu, H., Gao, M., Hou, H., & Wang, C. (2018).
Machine learning and deep learning methods for cybersecurity. IEEE Access, 6, 35365–35381.
Zhang, X. (2021). Application of data mining and machine learning in management accounting
information system. Journal of Applied Science and Engineering, 24(5), 813–820.