The modeling of genetic and Tabu search algorithm based BP neural network in the risk analysis of investment

There are many parameters which affect the economic benefit indicators for evaluation and they are all uncertain. Because of the parameters' uncertainty, evaluation indicators values of economic benefit in project investment are also uncertain. In these years, more and more BP Neural network based models are established to evaluate the risk of investment for its characters of wide information distribution and strong fault tolerance. However, the internal defect of local minima problem in this model will greatly discount the effects of evaluation if the training data are not proposed selected. To solve this problem of traditional BP neural network, an optimization algorithm that combines the advantages of genetic algorithm (GA) and Tabu Search (TS) is proposed. The training process is divided into two phases in order to search promising parameters of the BP neural network in favor of better evaluation and analysis of the investment. In the first phase, the initial parameters are first searched by GA algorithm by taking advantage of its various searches of the solutions, and the best parameters are selected by Tabu search algorithm in the second phases. After calculation of practical data, it proves that the new algorithm has better convergence rate and predictive accuracy, which make the evaluation and risk analysis of the investment faster and more accurate.