粗糙集BP神经网络在房地产价格预测中的应用

Application on Real Estate Price Prediction Based on RS and PBNN

研究房地产价格准确预测问题。由于房地产价格影响因子间信息严重冗余,受到社会上多种因素的影响。传统预测方法不能消除因子间的冗余信息,导致学习时间长、预测精度低。为了提高房地产价格的预测精度,提出一种粗糙集理论BP神经网络的房地产价格预测模型(RS-BPNN)。RS-BPNN模型首先采用粗糙集理论消除房地产价格因子间冗余信息,提取重要因子,然后采用非线性预测能力非常强的BP神经网络对处理后的数据进行学习建模,用建立好的模型对房地产价格进行预测。仿真结果表明,RS-BPNN房地产价格预测速度比传统预测方法快,预测精度更高,说明RS-BPNN的预测结果可以为政策制定者和房地商及买房提供参考。

Due to the influence of many factors in the real estate prices,redundant information exist among various factors,and the traditional neural network training time is to long.We use rough set theory in real estate price prediction,and put forward the real estate price predict model based on rough set theory combined with BP neural network.Rough sets theory has stronger ability to deal with uncertain information,which can insure the prediction accuracy,remove the redundant information and retain the necessary elements in the real estate price forecast,reduce the number of the input neuron effectively and simplify the structure of BP neuron networks.The empirical research shows that,compared with traditional neural network,the combination model＇s operating speed is faster,the forecasting accuracy is better and valid.