摘要
介绍了广义回归神经网络的原理,利用深圳市景田区房地产数据建立价格预测模型。为减少房地产估价过程中人为因素的影响,先从选取的12种影响因子分别对房地产价格做相关性分析,然后选出影响因子显著的6个作为网络输入。通过与传统的BP神经网络作比较,优化网络输入后的GRNN的预测效果更好,更有利于房地产市场的估计分析。
The theory of the Generalized Regression Neural Network is introduced in this paper. Then a price forecasting model is established using the real estate data of Jingtian dis- trict, Shenzhen.To reduce effect of human factors during the re- al estate appraisal process, a Correlation Analysis between twelve influence factors and real estate prices is calculated. And then six most significant impact factors are picked out as the network input. The GRNN with optimized network input has a better forecast result, when compared with the tradition- al BP neural network. And it is more conducive to the estimate analysis in the real estate markets.
出处
《测绘地理信息》
2014年第2期27-30,共4页
Journal of Geomatics
基金
国家自然科学基金资助项目(41174010
41374011)
