Comparing the Accuracy of Earnings Management Forecast Using Ant Colony Optimization Algorithm and Bacteria Foraging Algorithm

The present study is aimed to assess whether earnings management can be discovered on the basis of Machine Learning methods, so models based on Machine Learning (Ant Colony Optimization Algorithm and Bacteria Foraging) are applied to forecast earnings management. To do this, 143 firms listed in Tehran Stock Exchange are examined over a period from 2009 to 2013. Furthermore, Particle Swarm Optimization (PSO) is utilized in order to distinguish significant variables of earnings management and finally, earnings management is forecasted through the application of Matlab Software. Findings achieved from the fitness of Bacteria Foraging and Ant Colony Optimization algorithms indicates that these two algorithms are capable of forecasting earnings management with the accuracy of %98. Results show that Ant Colony Optimization model is more successful (error: %0.97) than Bacteria Foraging (error: %1.19) in earnings management forecasting.