随机模糊神经网络在公司收益预测中的应用

Stochastic Fuzzy Neural Network and It's Application in the Forecasting of Corporation's Earnings

系统介绍了随机模糊神经网络(SFNN)并将其用于上市公司每股收益的分类和预测,以解决一般的预测方法没有考虑到输入输出数据含有“噪声”的现实问题。文中选取了300家当前正在上海或深圳证券交易所上市的公司作为样本,用随机模糊神经网络对其每股收益的分类和预测进行了仿真研究,并将仿真结果与用模糊神经网络(FNN)的仿真结果进行了对比,结果显示用随机模糊神经网络的仿真效果较好。这对于投资者合理把握投资机会,正确投资以获得更高的收益有着一定的现实意义。

The paper systematically introduces the stochastic fuzzy neural network(SFNN) and applies it to the classification and forecasting of public corporation抯 earnings per share to settle the noise problem that the common forecasting methods have not considered. We choose 300 public corporations from Shanghai and Shenzhen stock markets as samples, and use the stochastic fuzzy neural network to simulate the classification and forecasting of earnings per share. We also use the fuzzy neural network to simulate and compare these two results. It shows that the simulation result of the stochastic fuzzy neural network is better. This provides some practical meanings for the investors to grasp the investing chance and make right investment decision to receive high earnings.