SIG ASYS - Accounting Info Systems
Loading...
Paper Type
Complete
Paper Number
1230
Description
When examining a firm’s financial statements, independent auditors seek to render opinions on their fairness, accuracy, presence of fraud, and going concern, among others. This research focuses on the going concern, and the ability to predict when the going concern is flagged based on an array of accounting measures. It seeks to determine a parsimonious set of measures that can accurately predict when the going concern is raised, when using a linear kernel support vector machine for prediction. A genetic algorithm is employed to effectively reduce the set of measures without compromising accuracy of prediction. Using data from audits of public firms, a parsimonious model is created utilizing only 8 measures from a set of 35 available measures. The model exhibits 98.6% accuracy, and outperforms several other machine learning techniques.
Recommended Citation
Yang, Tao and Nazareth, Derek, "Genetic Algorithm-based Feature Selection for Auditing Decisions" (2022). AMCIS 2022 Proceedings. 1.
https://aisel.aisnet.org/amcis2022/sig_ais/sig_ais/1
Genetic Algorithm-based Feature Selection for Auditing Decisions
When examining a firm’s financial statements, independent auditors seek to render opinions on their fairness, accuracy, presence of fraud, and going concern, among others. This research focuses on the going concern, and the ability to predict when the going concern is flagged based on an array of accounting measures. It seeks to determine a parsimonious set of measures that can accurately predict when the going concern is raised, when using a linear kernel support vector machine for prediction. A genetic algorithm is employed to effectively reduce the set of measures without compromising accuracy of prediction. Using data from audits of public firms, a parsimonious model is created utilizing only 8 measures from a set of 35 available measures. The model exhibits 98.6% accuracy, and outperforms several other machine learning techniques.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
SIG ASYS