A Learning Automaton Based Algorithm for Bankruptcy Prediction of Acceptable Firms within Power and Energy Exchange

In today's world, insurance of productive capital investment and reducing economical risk causes more fundraising and therefore the greatest economic boom cycle. One way to arrive capital investment security is to predict bankruptcy of a business unit. As the Iranian power and energy stock is going to start working by 2012, providing suitable advices to investors would be a priority. This paper proposes a new solution approach for bankruptcy prediction of the Iranian power and energy industries. To do so, an evolutionary algorithm premised on Learning Automata is employed and adapted to the problem. Two sets of firms related to power and energy industries that are listed on the Tehran Stock Exchange (TSE) are selected as the training and test data, respectively. The developed algorithm is conducted on both train and test data, and the efficiency of the proposed method is evaluated via several scenarios. It was practically seen in simulations that the learning automata based algorithm could achieve an accuracy of 91% and 88% over the train and test data, respectively. Besides these, the same data sets are also conducted by other methods such as MDA and Logit, and the obtained results are compared with reality. The yielded results prove the accuracy as well as the efficiency of the proposed solution technique