Title: Forecasting audit opinion based on multilevel perceptron neural network model using one-goal particle swarm optimisation
Authors: Fahimeh Bahrami; Javad Rezazadeh; Fatemeh Sarraf
Addresses: Department of Accounting, Raja University, Qazvin, Iran ' Department of Accounting, Faculty of Social Sciences, Imam Khomeini International University, Qazvin, Iran ' Department of Management and Accounting, South Tehran Branch, Islamic Azad University, Tehran, Iran
Abstract: Audit opinion is not necessarily absolute due to constraints inherent in any audit work, including sampling, different convincing degrees of audit evidence, accounting system and internal control characteristics, as well as inherent constraints of accounting measurement. The information contained in audit reports is used to forecast corporate bankruptcy; therefore, the auditor's opinion influences the views of users of financial statements. Financial status of a business can be assessed using financial ratios. The circumstances considered by an auditor in commenting on financial distress including financials (in particular, short-term liquidity, ability to repay and debt tolerability), and operations (profitability and ability to generate cash from operations). This study uses an integrated ANN-PSO to forecast audit opinion. The results show that ANN-PSO is more useful, because PSO algorithm can optimise hidden layers and neurons in ANN simultaneously. Finally, ANN-PSO outperforms ANN.
Keywords: forecast; audit opinion; neural network; multilayer perceptron; particle swarm optimisation; PSO; optimisation algorithm.
International Journal of Management Practice, 2020 Vol.13 No.1, pp.86 - 102
Received: 27 Apr 2018
Accepted: 20 Jun 2018
Published online: 06 Dec 2019 *