Johan Perols - Academia.edu (original) (raw)
Papers by Johan Perols
Journal of Forensic Accounting Research, 2021
Accounting firms are making significant investments in audit data analytics technologies to moder... more Accounting firms are making significant investments in audit data analytics technologies to modernize their audit services and the audit profession is believed to be on the verge of a transformation (BDO 2016; Deloitte 2016; EY 2015; Forbes Insights 2015; PwC 2015). In particular, the firms are emphasizing newer technologies such as interactive data visualization (BDO 2016; Deloitte 2016; PwC 2016) and they are increasingly expecting students to have data analytics skills (Forbes Insights 2015; PwC 2015). In this case you take on the role of Bryan, an audit senior assigned to Acme Company, who has been tasked with using interactive data visualization to gain an understanding of Acme's sales and perform an initial evaluation of two fraud risks identified during a fraud brainstorming session. Bryan has been given a data file with over 250,000 financial transactions and five master tables that he needs to analyze using interactive data visualization.
Improved classification performance has practical real-world benefits ranging from improved effec... more Improved classification performance has practical real-world benefits ranging from improved effectiveness indetecting diseases to increased efficiency in identifying firms that are committing financial fraud. Multiclas-sifier combination (MCC) aims to improve classification performance by combining the decisions of multiple individual classifiers. In this paper, we present information market-based fusion (IMF), a novel multiclassifier combiner method for decision fusion that is based on information markets. In IMF, the individual classifiers are implemented as participants in an information market where they place bets on different object classes. The reciprocals of the market odds that minimize the difference between the total betting amount and the potential payouts for different classes represent the MCC probability estimates of each class being the true object class. By using a market-based approach, IMF can adjust to changes in base-classifier performance without requiring offl...
Overview In this project, we are designing and experimentally testing a novel multi-classifier co... more Overview In this project, we are designing and experimentally testing a novel multi-classifier combination combiner method for decision fusion. Our combiner method, referred to as Information Market based Fusion (IMF), is based on parimutual betting markets. Using computational experiments involving 17 datasets from the UCI Machine Learning Repository and 22 different base-classifiers from Weka, we have initial results that show that IMF significantly outperforms Majority (MAJ), Average (AVG) and Weighted Average (WAVG) combiner method schemes. We are currently performing sensitivity analysis.
Information Market Based Decision Fusion 1 Johan Perols, Kaushal Chari, Manish Agrawal 2 Universi... more Information Market Based Decision Fusion 1 Johan Perols, Kaushal Chari, Manish Agrawal 2 University of South Florida, Tampa, Florida 33620 3 {jperols, kchari, magrawal}@coba.usf.edu 4 5 In this paper we present Information Market based Fusion (IMF), a novel multi-classifier combiner method for decision 6 fusion that is based on information markets. We compare the effectiveness of IMF to Majority, Average and Weighted 7 Average schemes using computational experiments involving 17 datasets from the UCI Machine Learning Repository 8 and 22 different base-classifiers from Weka. Collectively, the experimental results show that IMF outperforms the 9 other combiner methods. IMF furthermore adjusts to changes in base-classifier accuracy, provides incentives for the 10 base-classifiers to present truthful information, and does not require training or a static ensemble composition. 11 Additionally, in this paper we also introduce and test a novel dynamic cutoff selection algorithm. 12 13
In this paper, we present Information Market based Fusion (IMF), a novel, multi-classifier combin... more In this paper, we present Information Market based Fusion (IMF), a novel, multi-classifier combiner method for decision fusion that is based on information markets. IMF does not require training or a static ensemble composition, adjusts to changes in base-classifier accuracy, provides incentives for the base-classifiers to present truthful information, and integrates with existing multi-agent system (MAS) coordination mechanisms. We compare the effectiveness of two different IMF implementations to Majority (MAJ), Average (AVG), and Weighted Average (WAVG) schemes, using computational experiments involving 16 datasets from the UCI Machine Learning Repository and 20 different base-classifiers from Weka.
Academy of Management Proceedings, 2012
This study focuses on how inter-organizational collaboration contributes to build innovative capa... more This study focuses on how inter-organizational collaboration contributes to build innovative capability in enterprises. Using original data collected on collaborative relationships from Montreal-ba...
Journal of Information Systems, 2012
We extend continuous assurance research by proposing a novel continuous assurance architecture gr... more We extend continuous assurance research by proposing a novel continuous assurance architecture grounded in information fusion research. Existing continuous assurance architectures focus primarily on methods of monitoring assurance clients' systems to detect anomalous activities and have not addressed the question of how to process the detected anomalies. Consequently, actual implementations of these systems typically detect a large number of anomalies, with the resulting information overload leading to suboptimal decision making due to human information processing limitations. The proposed architecture addresses these issues by performing anomaly detection, aggregation and evaluation. Within the proposed architecture, artifacts developed in prior continuous assurance, ontology, and artificial intelligence research are used to perform the detection, aggregation and evaluation information fusion tasks. The architecture contributes to the academic continuous assurance literature and has implications for practitioners involved in the development of more robust and useful continuous assurance systems.
Academy of Management Proceedings, 2012
Prior literature emphasizes the trade-off between differentiation and integration as an important... more Prior literature emphasizes the trade-off between differentiation and integration as an important structural component of organizational ambidexterity. However, empirical research on the relationsh...
To Becca who provided support (in many ways), encouragement and motivation, helped me with my ide... more To Becca who provided support (in many ways), encouragement and motivation, helped me with my ideas, and believed in me more than I sometimes did; and to family and friends for providing the motivation for completing this dissertation. To the faculty, administrators and fellow Ph.D. students in the information systems and decision sciences department and the accounting department, thank you for all the support and for creating an excellent learning environment.
Auditing a Journal of Practice Theory, Mar 21, 2010
SUMMARY This study compares the performance of six popular statistical and machine learning model... more SUMMARY This study compares the performance of six popular statistical and machine learning models in detecting financial statement fraud under different assumptions of misclassification costs and ratios of fraud firms to nonfraud firms. The results show, somewhat surprisingly, that logistic regression and support vector machines perform well relative to an artificial neural network, bagging, C4.5, and stacking. The results also reveal some diversity in predictors used across the classification algorithms. Out of 42 predictors examined, only six are consistently selected and used by different classification algorithms: auditor turnover, total discretionary accruals, Big 4 auditor, accounts receivable, meeting or beating analyst forecasts, and unexpected employee productivity. These findings extend financial statement fraud research and can be used by practitioners and regulators to improve fraud risk models. Data Availability: A list of fraud companies used in this study is available from the author upon request. All other data sources are described in the text.
SSRN Electronic Journal, 2000
Improved classification performance has practical real-world benefits ranging from improved effec... more Improved classification performance has practical real-world benefits ranging from improved effectiveness in detecting diseases to increased efficiency in identifying firms that are committing financial fraud. Multiclassifier combination (MCC) aims to improve classification ...
Management Science, 2009
Improved classification performance has practical real-world benefits ranging from improved effec... more Improved classification performance has practical real-world benefits ranging from improved effectiveness in detecting diseases to increased efficiency in identifying firms that are committing financial fraud. Multiclassifier combination (MCC) aims to improve classification ...
Journal of Operations Management, 2013
ABSTRACT Supplier selection decisions are characterized by a high degree of uncertainty. We draw ... more ABSTRACT Supplier selection decisions are characterized by a high degree of uncertainty. We draw upon the behavioral operations management and decision-making literatures to examine factors that lead to the adoption of procedural rationality as a decision strategy. In addition, we emphasize the effect of procedural rationality on decision-makers’ perceived uncertainty and subsequent supplier decision performance. Our structural equation model with cross-country survey data from 461 respondents in the United States and China reveals that (i) organizational, situational, and personal antecedents significantly influence the use of procedural rationality, (ii) procedural rationality is effective in reducing uncertainty in supplier selection decisions, and (iii) the reduction in decision uncertainty improves supplier decision performance. We also emphasize contextual idiosyncrasies between China and the United States.
Journal of Forensic Accounting Research, 2021
Accounting firms are making significant investments in audit data analytics technologies to moder... more Accounting firms are making significant investments in audit data analytics technologies to modernize their audit services and the audit profession is believed to be on the verge of a transformation (BDO 2016; Deloitte 2016; EY 2015; Forbes Insights 2015; PwC 2015). In particular, the firms are emphasizing newer technologies such as interactive data visualization (BDO 2016; Deloitte 2016; PwC 2016) and they are increasingly expecting students to have data analytics skills (Forbes Insights 2015; PwC 2015). In this case you take on the role of Bryan, an audit senior assigned to Acme Company, who has been tasked with using interactive data visualization to gain an understanding of Acme's sales and perform an initial evaluation of two fraud risks identified during a fraud brainstorming session. Bryan has been given a data file with over 250,000 financial transactions and five master tables that he needs to analyze using interactive data visualization.
Improved classification performance has practical real-world benefits ranging from improved effec... more Improved classification performance has practical real-world benefits ranging from improved effectiveness indetecting diseases to increased efficiency in identifying firms that are committing financial fraud. Multiclas-sifier combination (MCC) aims to improve classification performance by combining the decisions of multiple individual classifiers. In this paper, we present information market-based fusion (IMF), a novel multiclassifier combiner method for decision fusion that is based on information markets. In IMF, the individual classifiers are implemented as participants in an information market where they place bets on different object classes. The reciprocals of the market odds that minimize the difference between the total betting amount and the potential payouts for different classes represent the MCC probability estimates of each class being the true object class. By using a market-based approach, IMF can adjust to changes in base-classifier performance without requiring offl...
Overview In this project, we are designing and experimentally testing a novel multi-classifier co... more Overview In this project, we are designing and experimentally testing a novel multi-classifier combination combiner method for decision fusion. Our combiner method, referred to as Information Market based Fusion (IMF), is based on parimutual betting markets. Using computational experiments involving 17 datasets from the UCI Machine Learning Repository and 22 different base-classifiers from Weka, we have initial results that show that IMF significantly outperforms Majority (MAJ), Average (AVG) and Weighted Average (WAVG) combiner method schemes. We are currently performing sensitivity analysis.
Information Market Based Decision Fusion 1 Johan Perols, Kaushal Chari, Manish Agrawal 2 Universi... more Information Market Based Decision Fusion 1 Johan Perols, Kaushal Chari, Manish Agrawal 2 University of South Florida, Tampa, Florida 33620 3 {jperols, kchari, magrawal}@coba.usf.edu 4 5 In this paper we present Information Market based Fusion (IMF), a novel multi-classifier combiner method for decision 6 fusion that is based on information markets. We compare the effectiveness of IMF to Majority, Average and Weighted 7 Average schemes using computational experiments involving 17 datasets from the UCI Machine Learning Repository 8 and 22 different base-classifiers from Weka. Collectively, the experimental results show that IMF outperforms the 9 other combiner methods. IMF furthermore adjusts to changes in base-classifier accuracy, provides incentives for the 10 base-classifiers to present truthful information, and does not require training or a static ensemble composition. 11 Additionally, in this paper we also introduce and test a novel dynamic cutoff selection algorithm. 12 13
In this paper, we present Information Market based Fusion (IMF), a novel, multi-classifier combin... more In this paper, we present Information Market based Fusion (IMF), a novel, multi-classifier combiner method for decision fusion that is based on information markets. IMF does not require training or a static ensemble composition, adjusts to changes in base-classifier accuracy, provides incentives for the base-classifiers to present truthful information, and integrates with existing multi-agent system (MAS) coordination mechanisms. We compare the effectiveness of two different IMF implementations to Majority (MAJ), Average (AVG), and Weighted Average (WAVG) schemes, using computational experiments involving 16 datasets from the UCI Machine Learning Repository and 20 different base-classifiers from Weka.
Academy of Management Proceedings, 2012
This study focuses on how inter-organizational collaboration contributes to build innovative capa... more This study focuses on how inter-organizational collaboration contributes to build innovative capability in enterprises. Using original data collected on collaborative relationships from Montreal-ba...
Journal of Information Systems, 2012
We extend continuous assurance research by proposing a novel continuous assurance architecture gr... more We extend continuous assurance research by proposing a novel continuous assurance architecture grounded in information fusion research. Existing continuous assurance architectures focus primarily on methods of monitoring assurance clients' systems to detect anomalous activities and have not addressed the question of how to process the detected anomalies. Consequently, actual implementations of these systems typically detect a large number of anomalies, with the resulting information overload leading to suboptimal decision making due to human information processing limitations. The proposed architecture addresses these issues by performing anomaly detection, aggregation and evaluation. Within the proposed architecture, artifacts developed in prior continuous assurance, ontology, and artificial intelligence research are used to perform the detection, aggregation and evaluation information fusion tasks. The architecture contributes to the academic continuous assurance literature and has implications for practitioners involved in the development of more robust and useful continuous assurance systems.
Academy of Management Proceedings, 2012
Prior literature emphasizes the trade-off between differentiation and integration as an important... more Prior literature emphasizes the trade-off between differentiation and integration as an important structural component of organizational ambidexterity. However, empirical research on the relationsh...
To Becca who provided support (in many ways), encouragement and motivation, helped me with my ide... more To Becca who provided support (in many ways), encouragement and motivation, helped me with my ideas, and believed in me more than I sometimes did; and to family and friends for providing the motivation for completing this dissertation. To the faculty, administrators and fellow Ph.D. students in the information systems and decision sciences department and the accounting department, thank you for all the support and for creating an excellent learning environment.
Auditing a Journal of Practice Theory, Mar 21, 2010
SUMMARY This study compares the performance of six popular statistical and machine learning model... more SUMMARY This study compares the performance of six popular statistical and machine learning models in detecting financial statement fraud under different assumptions of misclassification costs and ratios of fraud firms to nonfraud firms. The results show, somewhat surprisingly, that logistic regression and support vector machines perform well relative to an artificial neural network, bagging, C4.5, and stacking. The results also reveal some diversity in predictors used across the classification algorithms. Out of 42 predictors examined, only six are consistently selected and used by different classification algorithms: auditor turnover, total discretionary accruals, Big 4 auditor, accounts receivable, meeting or beating analyst forecasts, and unexpected employee productivity. These findings extend financial statement fraud research and can be used by practitioners and regulators to improve fraud risk models. Data Availability: A list of fraud companies used in this study is available from the author upon request. All other data sources are described in the text.
SSRN Electronic Journal, 2000
Improved classification performance has practical real-world benefits ranging from improved effec... more Improved classification performance has practical real-world benefits ranging from improved effectiveness in detecting diseases to increased efficiency in identifying firms that are committing financial fraud. Multiclassifier combination (MCC) aims to improve classification ...
Management Science, 2009
Improved classification performance has practical real-world benefits ranging from improved effec... more Improved classification performance has practical real-world benefits ranging from improved effectiveness in detecting diseases to increased efficiency in identifying firms that are committing financial fraud. Multiclassifier combination (MCC) aims to improve classification ...
Journal of Operations Management, 2013
ABSTRACT Supplier selection decisions are characterized by a high degree of uncertainty. We draw ... more ABSTRACT Supplier selection decisions are characterized by a high degree of uncertainty. We draw upon the behavioral operations management and decision-making literatures to examine factors that lead to the adoption of procedural rationality as a decision strategy. In addition, we emphasize the effect of procedural rationality on decision-makers’ perceived uncertainty and subsequent supplier decision performance. Our structural equation model with cross-country survey data from 461 respondents in the United States and China reveals that (i) organizational, situational, and personal antecedents significantly influence the use of procedural rationality, (ii) procedural rationality is effective in reducing uncertainty in supplier selection decisions, and (iii) the reduction in decision uncertainty improves supplier decision performance. We also emphasize contextual idiosyncrasies between China and the United States.