摘要
标准的BP神经网络算法在训练过程中没有具体和明确的理论指导,容易陷入局部极小值以及泛化能力与训练的矛盾等问题,因此提出粒子群(PSO)算法改进BP神经网络模型,选择资产、收入、债务、经济需求、时间观念五个输入变量预测普通投资者的年龄,并与标准的BP算法比较分析.实验结果表明:PSO-BP算法比标准的BP算法在评价指标上具有更低的平均绝对误差MSE、均方误差MAE、均方根误差RMSE以及平均绝对百分比误差MAPE,PSO-BP算法能够更精确地预测到普通投资者的年龄值,具有更小的误差值,这对投资者的投资导向研究具有一定的参考价值.
Abstract
The standard BP neural network algorithm lacks specific and clear theoretical guidance during the training process,which can easily lead to problems such as local minima and the contradiction between generalization ability and training.There-fore,a particle swarm optimization(PSO)algorithm is proposed to improve the BP neural network model,selecting five input vari-ables:asset,income,debt,economic demand,and time to predict the age of ordinary investors,then made a comparative analysis with the standard BP algorithm,The experimental results show that the PSO-BP algorithm has lower mean square error(MSE),mean absolute error(MAE),root mean square error(RMSE),and mean absolute percentage error(MAPE)in evaluation indicators compared to the standard BP algorithm.The PSO-BP algorithm can more accurately predict the age value of ordinary investors with smaller error values,which has certain reference value for investment orientation research of investors.
NETL
NSTL
万方数据