The Essentials of Machine Learning in Finance and Accounting

Cover
Half Title
Series Page
Title Page
Copyright Page
Table of Contents
List of figures
List of tables
Notes on contributors
1 Machine learning in finance and accounting
	1.1 Introduction
	1.2 Motivation
	1.3 Brief overview of chapters
	References
2 Decision trees and random forests
	2.1 Introduction
	2.2 Classification trees
		2.2.1 Impurity and binary splitting
			2.2.1.1 Specification of the impurity function
			2.2.1.2 Labeling the leaves
			2.2.1.3 Tree size and stopping rules
		2.2.2 Performance estimation
			2.2.2.1 Resubstitution estimate
			2.2.2.2 Test-sample estimate
	2.3 Regression trees
		2.3.1 Regression
		2.3.2 Performance assessment and optimal size of the tree
			2.3.2.1 Resubstitution estimate of MSE(T)
			2.3.2.2 Test-sample estimate of MSE(T)
	2.4 Issues common to classification and regression trees
		2.4.1 Surrogate splits
			2.4.1.1 Handling of missing values
			2.4.1.2 Ranking of input variables
			2.4.1.3 Input combination
		2.4.2 Advantages and disadvantages of decision trees
	2.5 Random forests
		2.5.1 Prediction error bias-variance decomposition
		2.5.2 Bias-variance decomposition for randomized trees ensembles
		2.5.3 From trees ensembles to random forests
		2.5.4 Partial dependence function
	2.6 Forecasting bond returns using macroeconomic variables
	2.7 Default prediction based on accountancy data
	2.8 Appendix: R source codes for the applications in this chapter
		2.8.1 Application to US BofA index
		2.8.2 SME default risk application
	References
3 Improving longevity risk management through machine learning
	3.1 Introduction
	3.2 The mortality models
	3.3 Modeling mortality with machine learning
	3.4 Numerical application
		3.4.1 Mortality models by comparison: an empirical analysis
		3.4.2 Longevity management for life insurance: sample cases
	3.5 Conclusions
	3.6 Appendix
	Note
	References
4 Kernel switching ridge regression in business intelligence systems
	4.1 Introduction
	4.2 Method
		4.2.1 Switching regression
		4.2.2 Switching ridge regression
		4.2.3 Dual form of the ridge regression
		4.2.4 Basic notion of kernel methods
		4.2.5 Alternative derivation to use ridge regression in the feature space
		4.2.6 Kernel ridge regression
		4.2.7 Kernel ridge regression: duality
		4.2.8 Kernel switching ridge regression
	4.3 Experimental results
		4.3.1 Simulation
		4.3.2 Application in business intelligence
	4.4 Discussion
	4.5 Conclusion and future research
	4.6 Appendix: Kernel switching ridge regression: an R code
	References
5 Predicting stock return volatility using sentiment analysis of corporate annual reports
	5.1 Introduction
	5.2 Related literature
	5.3 Research methodology
		5.3.1 Financial data and indicators
		5.3.2 Textual data and linguistic indicators
		5.3.3 Machine learning methods
	5.4 Experimental results
	5.5 Conclusions
	Acknowledgments
	References
6 Random projection methods in economics and finance
	6.1 Introduction
	6.2 Dimensionality reduction
		6.2.1 Principal component analysis (PCA)
		6.2.2 Factor analysis
		6.2.3 Projection pursuit
	6.3 Random projection
		6.3.1 Johnson-Lindenstrauss lemma
		6.3.2 Projection matrices’ specification
	6.4 Applications of random projection
		6.4.1 A compressed linear regression model
		6.4.2 Tracking the S&P500 index
		6.4.3 Forecasting S&P500 returns
		6.4.4 Forecasting energy trading volumes
	6.5 Appendix: Matlab code
	Notes
	References
7 The future of cloud computing in financial services: a machine learning and artificial intelligence perspective
	7.1 Introduction
	7.2 The role of machine learning and artificial intelligence in financial services
	7.3 The enterprise data cloud
	7.4 Data contextuality: machine learning-based entity analytics across the enterprise
	7.5 Identifying Central Counterparty (CCP) risk using ABM simulations
	7.6 Systemic risk and cloud concentration risk exposures
	7.7 How should regulators address these challenges?
	Notes
	References
8 Prospects and challenges of using artificial intelligence in the audit process
	8.1 Introduction
		8.1.1 Background and relevant aspect of auditing
	8.2 Literature review
	8.3 Artificial intelligence in auditing
		8.3.1 Artificial intelligence
		8.3.2 Use of expert systems in auditing
		8.3.3 Use of neural network in auditing
	8.4 Framework for including AI in auditing
		8.4.1 Components
			8.4.1.1 AI strategy
			8.4.1.2 Governance
			8.4.1.3 Human factor
		8.4.2 Elements
			8.4.2.1 Cyber resilience
			8.4.2.2 AI competencies
			8.4.2.3 Data quality
			8.4.2.4 Data architecture and infrastructure
			8.4.2.5 Measuring performance
			8.4.2.6 Ethics
			8.4.2.7 Black box
	8.5 Transformation of the audit process
		8.5.1 Impact of digitalization on audit quality
		8.5.2 Impact of digitalization on audit firms
		8.5.3 Steps to transform manual audit operations to AI-based
	8.6 Applications of artificial intelligence in auditing – few examples
		8.6.1 KPMG
		8.6.2 Deloitte
		8.6.3 PwC
		8.6.4 Ernst and Young (EY)
		8.6.5 K.Coe Isom
		8.6.6 Doeren Mayhew
		8.6.7 CohnReznick
		8.6.8 The Association of Certified Fraud Examiners (ACFE)
	8.7 Prospects of an AI-based audit process in Bangladesh
		8.7.1 General aspects
		8.7.2 Audit firm specific aspects
		8.7.3 Business organization aspects
	8.8 Conclusion
	Bibliography
9 Web usage analysis: pillar 3 information assessment in turbulent times
	9.1 Introduction
	9.2 Related work
	9.3 Research methodology
	9.4 Results
	9.5 Discussion and conclusion
	Acknowledgements
	Disclosure statement
	References
10 Machine learning in the fields of accounting, economics and finance: the emergence of new strategies
	10.1 Introduction
	10.2 General overview on machine learning
	10.3 Data analysis process and main algorithms used
		10.3.1 Supervised models
		10.3.2 Unsupervised models
		10.3.3 Semi-supervised models
		10.3.4 Reinforcement learning models
	10.4 Machine learning uses: cases in the fields of economics, finance and accounting
		10.4.1 Algorithmic trading
		10.4.2 Insurance pricing
		10.4.3 Credit risk assessment
		10.4.4 Financial fraud detection
	10.5 Conclusions
	References
11 Handling class imbalance data in business domain
	11.1 Introduction
	11.2 Data imbalance problem
	11.3 Balancing techniques
		11.3.1 Random sampling-based method
		11.3.2 SMOTE oversampling
		11.3.3 Borderline-SMOTE
		11.3.4 Class weight boosting
	11.4 Evaluation metrics
	11.5 Case study: credit card fraud detection
	11.6 Conclusion
	References
12 Artificial intelligence (AI) in recruiting talents: recruiters’ intention and actual use of AI
	12.1 Introduction
	12.2 Theory and hypothesis development
		12.2.1 Technology anxiety and intentions to use
		12.2.2 Performance expectancy and intentions to use
		12.2.3 Effort expectancy and intentions to use
		12.2.4 Social influence and intention to use
		12.2.5 Resistance to change and intentions to use
		12.2.6 Facilitating conditions and intentions to use
		12.2.7 Behavioral intention to use and actual use
		12.2.8 Moderating effects of age status
	12.3 Research design
		12.3.1 Survey design
		12.3.2 Data collection procedure and participants’ information
		12.3.3 Measurement tools
		12.3.4 Results and hypotheses testing
			12.3.4.1 Analytical technique
			12.3.4.2 Measurement model evaluation
			12.3.4.3 Structural model evaluation
			12.3.4.4 Testing of direct effects
			12.3.4.5 Testing of moderating effects
	12.4 Discussion and conclusion
		12.4.1 Limitation of study and future research directions
	References
Index