Novel Financial Applications of Machine Learning and Deep Learning: Algorithms, Product Modeling, and Applications

235 Pages·2023·6.228 MB·English

by  Mohammad Zoynul Abedin, Petr Hajek

International Series in
Operations Research & Management Science
Mohammad Zoynul Abedin
Petr Hajek   Editors
Novel Financial
Applications
of Machine
Learning and Deep
Learning
Algorithms, Product Modeling, and
Applications
International Series in Operations Research &
Management Science
FoundingEditor
FrederickS.Hillier,StanfordUniversity,Stanford,CA,USA
Volume 336
SeriesEditor
Camille C. Price, Department of Computer Science, Stephen F. Austin State Uni-
versity,Nacogdoches,TX,USA
EditorialBoardMembers
Emanuele Borgonovo, Department of Decision Sciences, Bocconi University,
Milan,Italy
BarryL.Nelson,DepartmentofIndustrialEngineering&ManagementSciences,
NorthwesternUniversity,Evanston,IL,USA
BruceW.Patty,VeritecSolutions,MillValley,CA,USA
MichaelPinedo,SternSchoolofBusiness,NewYorkUniversity,NewYork,NY,
USA
RobertJ.Vanderbei,PrincetonUniversity,Princeton,NJ,USA
AssociateEditor
Joe Zhu, Foisie Business School, Worcester Polytechnic Institute, Worcester, MA,
USA
The book series International Series in Operations Researchand Management
Science encompasses the various areas of operations research and management
science. Both theoretical and applied books are included. It describes current
advances anywhere intheworld thatareatthecuttingedgeofthefield.Theseries
is aimed especially at researchers, advanced graduate students, and sophisticated
practitioners.
Theseriesfeaturesthreetypesofbooks:
(cid:129)Advancedexpositorybooksthatextendandunifyourunderstandingofpartic-
ularareas.
(cid:129)Researchmonographsthatmakesubstantialcontributionstoknowledge.
(cid:129) Handbooks that define the new state of the art in particular areas. Each
handbook will be edited by a leading authority in the area who will organize a
team of experts on various aspects of the topic to write individual chapters. A
handbook may emphasize expository surveys or completely new advances (either
researchorapplications)oracombinationofboth.
Theseriesemphasizesthefollowingfourareas:
MathematicalProgramming: Including linearprogramming, integerprogram-
ming, nonlinear programming, interior point methods, game theory, network opti-
mization models, combinatorics, equilibrium programming, complementarity
theory, multiobjective optimization, dynamic programming, stochastic program-
ming,complexitytheory,etc.
Applied Probability: Including queuing theory, simulation, renewal theory,
Brownian motion and diffusion processes, decision analysis, Markov decision
processes, reliability theory, forecasting, other stochastic processes motivated by
applications,etc.
ProductionandOperationsManagement:Includinginventorytheory,produc-
tion scheduling, capacity planning, facility location, supply chain management,
distributionsystems,materialsrequirementsplanning,just-in-timesystems,flexible
manufacturing systems, design of production lines, logistical planning, strategic
issues,etc.
Applications of Operations Research and Management Science: Including
telecommunications,healthcare,capitalbudgetingandfinance,economics,market-
ing, public policy, military operations research, humanitarian relief and disaster
mitigation,serviceoperations,transportationsystems,etc.
ThisbookseriesisindexedinScopus.
(cid:129)
Mohammad Zoynul Abedin Petr Hajek
Editors
Novel Financial Applications
of Machine Learning
and Deep Learning
Algorithms, Product Modeling,
and Applications
Editors
MohammadZoynulAbedin PetrHajek
DepartmentofFinance,Performanceand FacultyofEconomicsandAdministration
Marketing UniversityofPardubice
TeessideUniversityInternationalBusiness Pardubice,CzechRepublic
School,TeessideUniversity
Middlesbrough,UK
ISSN0884-8289 ISSN2214-7934 (electronic)
InternationalSeriesinOperationsResearch&ManagementScience
ISBN978-3-031-18551-9 ISBN978-3-031-18552-6 (eBook)
https://doi.org/10.1007/978-3-031-18552-6
©TheEditor(s)(ifapplicable)andTheAuthor(s),underexclusivelicensetoSpringerNatureSwitzerland
AG2023
Thisworkissubjecttocopyright.AllrightsaresolelyandexclusivelylicensedbythePublisher,whether
thewholeorpartofthematerialisconcerned,specificallytherightsoftranslation,reprinting,reuseof
illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by
similarordissimilarmethodologynowknownorhereafterdeveloped.
Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication
doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
The publisher, the authors, and the editorsare safeto assume that the adviceand informationin this
bookarebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsor
theeditorsgiveawarranty,expressedorimplied,withrespecttothematerialcontainedhereinorforany
errorsoromissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardtojurisdictional
claimsinpublishedmapsandinstitutionalaffiliations.
ThisSpringerimprintispublishedbytheregisteredcompanySpringerNatureSwitzerlandAG
Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland
Preface
TheNovelFinancialApplicationsofMachineLearningandDeepLearning:Algo-
rithms, Product Modelling, and Applications presents the state of the art of the
application of machine learning (ML) and deep learning (DL) in the domain of
finance. We will present a combination of empirical evidence to diverse fields of
financesothatthisbookisusefultoacademics,practitioners,andpolicymakerswho
arelookingtotrainnovelandthemostadvancedmachinelearningclassifiers.Thus,
the purpose of this book is to provide a broad area of applications to different
financial assets and markets. Furthermore, from an extensive literature assessment,
it is evident that there are no existing textbooks that narrate ML and DL to unlike
areasoffinanceortoanextensiverangeofproductsandmarkets.
Businessrisk and uncertaintycertainly arethetoughestchallengeinthe finance
domainfacedbymanyresearchersandmanagers.Suchuncertaintytherebyinitiates
an unavoidable risk factor, which is a fundamental element of financial theory. To
the best of our knowledge, the financial domain has not been a focused subject-
matterforgoodMLrelatedbooks.Thereisalsoascarcityofinformationabouthow
financial enterprises supervise crisis events and achieve turnaround. In order to fix
the multifarious nature of the financial problem, this edited book advocates inter-
disciplinaryapproachesbasedonmachinelearning.
Machine learning is involved in the analysis of large and multiple feature
instances. It principally refers to acquiring knowledge and intelligence (by a com-
puterprogram)fromaprocessedtrainingexampleforgeneratingpredictions.Itdeals
with computationallyintensivetechniques,suchascluster analysis, dimensionality
reduction,andsupportvectoranalysis.Itisprincipallytheareaofcomputerscience
andisalreadyfrequentlyappliedinsocialsciences,financeandbanking,marketing
research, operations research, and applied sciences. Moreover, computational
finance is a domain of applied computer science that is concerned with practical
issues in finance. It may be characterized as the study of features, instances, and
learning algorithms applied in finance. It is an interdisciplinary area that integrates
computational tools with numerical finance. Furthermore, computational finance
applies arithmetical proofs that can be fitted to economic experiments, thereby
v
vi Preface
contributingtotheadvancementoffinancialdatamodelingtechniquesandsystems.
Thesecomputationaltechniquesareutilizedinfinancialriskmanagement,corporate
bankruptcy prediction, stock price prediction, and portfolio management. Finally,
thisproposedtextbookcouldplayanimportantroleinfinancialdatalearning.
Besides, this volume will be a basis for empirical and theoretical practices. The
empirical experiments aim to minimize financial risk and uncertainty by covering
andfittingthemostadvancedandnovelmachinelearningalgorithms.Moreover,it
generates academic literature as well as financial product and finance modeling
inferencestowardcustomercreditriskassessment,datamining,patternrecognition,
bankruptcyprediction,andsoon.Tobespecific,thevolumeisbroadlydividedinto
three parts, with the first set of chapters focusing on the recent trend and issues of
financial technology (FinTech). The second set of chapters comprises empirical
essays on the prediction and forecasting financial risk by applying ML and DL
tools and techniques. The third set of chapters combines empirical evidence of
financial time-series data forecasting. The volume ends with a set of emerging
technologiesinfinancialeducationandhealthcareandtheirempiricalapplications.
Part 1: Recent Developments in FinTech
ThefirstpartpresentsfourchaptersonrecentdevelopmentinFinTech.
Chapter “FinTech Risk Management and Monitoring” focuses on risk manage-
ment and monitoring in FinTech. The recent emergence of financial technology
innovationsinthefinancialservicesandsomesignificantrisksareinvestigatedusing
the qualitative research method. Additionally, the appropriate way to mitigate the
risk is discussed in this chapter. Besides this objective, this chapter discusses the
major risk behind the rapid development of fintech and the steps for fintech risk
management.Thefourkeyregulatorytechniquesthathaveimportantapplicationsin
FinTech management and monitoring are added, and, finally, the chapter summa-
rizesthemainchallengesofFinTechriskmanagement.
Chapter “Digital Transformation of Supply Chain with Supportive Culture in
BlockchainEnvironment”explorestheinfluenceofblockchainonthedigitaltrans-
formation of Supply Chain Management (SCM). This chapter is also aimed to
determine the importance of supportive culture in the adoption of blockchain in
supply chains. The study findings indicate that the digitalization of supply chain
managementbyadoptingblockchaintechnologyispositivelycorrelatedwithorga-
nizationalprosperity.Thechapteralsoindicatesthatsupportivecultureiscrucialto
practicingblockchaintechnology.Thisstudysuggeststhatpolicymakersandstake-
holders ensure a supportive culture to establish a traceable, efficient, and effective
supplychain.
Chapter “Integration of Artificial Intelligence Technology in Management
Accounting Information System: An Empirical Study” conducts an empirical
studyontheintegrationofartificialintelligencetechnologyinmanagementaccount-
ing information systems. This study established an artificial neural network-based
Preface vii
model to predict management information and verify the accuracy of the model
using some real data. Five dimensions are considered to develop the model,
accounting analysis management system, accountingdecision supportsystem, per-
formance management information system, risk management information system,
andenvironmentmanagementinformationsystem.
The essentiality to analyze big data in accounting and finance is discussed in
Chap.“TheImpactofBigDataonAccountingPractices:EmpiricalEvidencefrom
Africa”. Evidence indicates that big data significantly impact accounting and
auditing accounting, utilizing the diversity of data volume, data variety, and data
velocity. Chapter “The Impact of Big Data on Accounting Practices: Empirical
Evidence from Africa” shows the impact of big data on accounting practices, and
thestudyareaisAfrica.Themaingoalofthischapteristoexploretheimpactsofbig
data on accounting using accountants in Nigeria. Multiple regression is used for
151responses, andsamplesarecollectedusingtherandomsamplingmethod.This
study proves that big data positively and significantly affect financial reporting,
performance measurement, corporate budgeting, audit evidence, risk management,
andfraudmanagement.Thisstudyhelpsaccountants,prospectiveaccountants,and
accountinggraduatesintheirstudies.
Part 2: Financial Risk Prediction Using Machine Learning
ThesecondpartcontainsfourchaptersthatdiscusstheapplicationsofMLandDL
approachestopredictandforecastfinancialrisk.
Chapter “Using Outlier Modification Rule for Improvement of the Performance
of Classification Algorithms in the Case of Financial Data” discusses how to
improve classifier performance by mining and modifying outliers of financial
datasets. This chapter offers insights into the Financial Decision Support System
for financial decision makers. This study employs four distinct classification algo-
rithms such as linear discriminant analysis, k-nearest neighbor, naïve Bayes, and
supportvectormachineforbothoriginalandmodifieddatasetstodetectcreditcard
fraud. The study’s findings show that the classifiers perform better on modified
datasetsthanonoriginalcreditcarddatasets.
Chapter “Default Risk Prediction Based on Support Vector Machine and Logit
SupportVectorMachine”isapredictiveanalysisofthemachinelearningalgorithm
fordefaultriskprediction.ThisstudyproposesaLogitSVM modelthathybridized
thetraditionalsupportvectormachinewithpopularlogisticregressiontoassessthe
credit default risk. The authors use real-world credit databases to validate the
probability and value of the proposed model. Type I error, type II error, and root
mean square error (RMSE) are used to evaluate the performance of the regressors.
Empirical findings show that the proposed hybrid model is superior to maximize
accuracy and minimize RMSE. This chapter helps stockholders develop a wide
varietyofapproachestopredictthecreditcustomers’defaultrisk.
viii Preface
Chapter “Predicting Corporate Failure Using Ensemble Extreme Learning
Machine” shows the corporate failure prediction using the Ensemble Extreme
Learning Machine. The claim is that the early-stage prediction of corporate failure
is essential for banks and financial institutions to solve financial decision-making
problems. Newly developed artificial intelligence technique Extreme Learning
Machine has an extremely fast learning classifier. To prove the superiority of this
method, the authors compare the result with four benchmark ensemble methods,
namelymultipleclassifiers,bagging,boosting,andrandomsubspace.Experimental
resultsonFrenchfirmsindicatedthatbaggedandboostedextremelearningmachines
showedthebest-improvedperformance.
Chapter “Assessing and Predicting Small Enterprises’ Credit Ratings: A
Multicriteria Approach” focuses on small enterprises; it assigns and predicts the
small enterprise’s credit rating using a multicriteria approach. In reality, small
enterprises have made it difficult for financial institutions such as commercial
banks to accurately determine the credit risk, creating salient loan difficulties due
toshorttime,highfrequency,urgentdemandforcredit,andasmallnumberoftheir
loans.Tosolvethisissue,thechapterdevelopsanewapproachforassessingcredit
riskinsmallenterprisesbycombininghigh-dimensionalattributereductionmethods
withfuzzyC-meanstogradethecreditratingsofenterprisesrequestingloans.
Part 3: Financial Time-Series Forecasting
The third part contains two chapters that explore empirical evidence of time-series
datamodeling.
Chapter“AnEnsembleLGBM(LightGradientBoostingMachine)Approachfor
Crude Oil Price Prediction” is on the prediction of crude oil prices. Every second
countswhengovernments,businesses,andindividualsneedtoknowwhatthefuture
of thecrude oil marketwill bring in terms ofpricing.Estimating thefuture cost of
crude oil is a crucial step toward building an economy that can last. In order to
effectivelypredictfuturecrudemarketprices,thisresearchwillusemachinelearning
andensemblelearningtechniques.Themodelusinglightgradientboosting(LGBM)
is proposed by the authors to predict the price of crude oil. By analyzing and
modeling the Brent time-series crude oil data, the accuracy and precision of our
predictorscanbeimproved.TheLBGMforecastiscomparedtothelassoregression,
randomforestregression,anddecisiontreeregressionmethods.Theresultsachieved
by the suggested model are quite similar to and better than those obtained by the
baseline model when measured using RMSE, mean absolute percentage error
(MAPE),meansquarederror(MSE),andmeanabsoluteerror(MAE).
Chapter “Model Development for Predicting the Crude Oil Price: Comparative
EvaluationofEnsembleandMachineLearningMethods”alsoshowstheprediction
ofcrudeoilpricesusingdifferentmethods.Thisstudyshowsacomparativestudyof
ensemble algorithms and machine learning algorithms to find the best forecasting
model.Thisresearch uses machine learning and an ensemblealgorithm to forecast
Preface ix
crudeoilprices,anditcomparestheefficacyofthreedifferentregressionmodels—
AdaBoost, Bagging Lasso, and Support Vector Regression—to conclude which is
the most suitable. Time-series data on crude oil prices are analyzed and used to
validate the forecasting model. The results of the various algorithms are compared
using an actual vs. anticipated curve. According to the results, the ensemble
AdaBoostmethodhassuperiorperformance.Themeansquareerror,meanabsolute
error,rootmeansquareerror,meanabsolutepercentageerror,variancescore,andR2
areusedtoverifytheoutcome.Thisresearchwillhelpthosewithastakeinthecrude
oilindustrydecideandcraftpoliciesbasedonprojectedfutureprices.
Part 4: Emerging Technologies in Financial Education
and Healthcare
The fourth part contains three chapters that explore the financial education and
healthcareissuesandtheiremergingtrends.
Chapter “Discovering the Role of M-Learning Among Finance Students: The
Future of Online Education” investigates the role of m-learning among finance
studentsandthefutureofonlinehighereducation.Thisstudyaimstofindthehidden
issuesofm-learninginfinancestudies.Thisstudyismainlyaqualitativeapproach,
andthefindingsshowthatdigitalizededucationprovidestheopportunityformajor
financestudentstoaccessfinancialmarketsusingtheInternetandgainpersonaland
professionalknowledgeinabetterwayratherthantraditionallearning.Italsoshows
thatm-learninghasasignificantpositiverelationshipwiththeeffectivenessofonline
education.Thisanalysishasasignificantimplicationforeducationpolicymakersand
practitioners.
Chapter “Exploring the Role of Mobile Technologies inHigher Education: The
Impactof Online Teaching on Traditional Learning” demonstrates how technolog-
ical evolvements derive the conduction of higher education, especially mobile
technology. This study also intended to detect the factors that attract pupils who
donotadoptanonlineeducationsystem.Aqualitativeapproachisusedtodetermine
the pros and cons of the technology-based education system in universities. The
authors reveal that the adoption of mobile technologies in academic education
enablesstudentstoaccessvaluableresourcesfreeofcostandeffortlessly,whichin
turn helps them to develop strong knowledge and understanding of their study
contents. This study opens up a new arena for research scholars to discover the
importanceofonlineeducationsystems.
Chapter“KnowledgeMiningfromHealthData:ApplicationofFeatureSelection
Approaches”assessedtheperformanceoffeatureselectiontechniquesinknowledge
mining of health datasets. This study compared seven popular knowledge mining
approaches on six popular Affymetrix and cDNA datasets. Employing a support
vectormachineclassifier,thestudydeterminedtheknowledgeminers’accuracyand
areaunderthecurvevalues.Thefindingofthischapterinformsthatthesimplelasso