一种前向型神经网络基准地价预测模型的研究

土地是国家重要资产,对地价的全面监测、分析和预测,有助于城市基准地价的快速更新和政府对土地资源的宏观控制。现阶段基准地价水平调整的预测模型有明显的不足。应用一种非线性预测方法:多层前向神经网络建立基准地价预测模型,采用时间预测序列。在介绍误差反向修正算法的基础上,通过预测实例计算,表明预测结果与实际情况吻合较好

Study on Benchmark Land-value Forecast by Means of Feedforward Neural Network

Soil is assets of nation, to watch、analyse and forecast the land-value.it use for government to macro-control, the now forecast model of benchmark land-value is insufficiency.This paper carries out a non-linear forecast method, multi-feed-forward neural network, is adopted as forcasting model and by list of time.Based on introducing discrepancy back-propagation algorithm.By practical calculation of forecast that the discrepancy back-propagation algorithm can get a better impact.