aM.A Student, Financial Management, University of Sistan and Baluchestan, Zahedan, Iran
bM.A in Accounting, University of Ferdosi, Mashad
cM.A in Accounting, University of Sistan and Baluchestan, Zahedan, Iran
Online published on 6 September, 2014.
The main aim of this research is to predict real earnings management using colonial competitive algorithm(ICA), Particle Swarm Optimization (PSO), support vector machine(SVM) and multi-layer Perceptron (MLP)network algorithm combined with Particle Swarm Optimization (PSO). For this purpose, the fourteen variables affecting earnings management were used as independent variables and real earnings management were used as dependent variables. Research sample include 113 firms for 7 years between 2008–2012. The results show that multi-layer Perceptron network is more efficient than other algorithms in real earnings management prediction. Eviews was used for earning management models and Matlab and SPSS used for their implementation.
Predicting real earning management, Colonial competitive algorithm, algorithm birds move, Support Vector Machine, Multi-layer Perceptron network