基于ARIMA-BP组合模型的房地产价格预测方法研究

针对使用单一预测模型存在数据特征提取不充分,预测精度不高的问题,提出了一种基于ARIMA-BP组合模型的房地产价格预测方法。结合ARIMA模型处理线性问题的优势以及BP神经网络模型在非线性问题上的优势,利用误差方差加权平均训练法训练出最佳权重的组合并建立组合模型对某市区房地产价格和趋势预测进行实证分析。理论分析和实验结果表明,所提两者的组合模型有效解决了不能充分提取数据特征,预测精度不理想的问题,比单一预测模型能获得更准确的预测效果。

Research on Real Estate Price Forecasting Method Based on ARIMA-BP Combination Model

In order to solve the problem of insufficient data feature extraction and low prediction accuracy in single prediction model, a real estate price prediction method based on ARIMA-BP combined model is proposed. Combined with the advantages of ARIMA model in dealing with linear problems and the advantages of BP neural network model in non-linear problems, the combination of the best weights is trained by using the weighted average training method of error variance and the combination model is established to make an empirical analysis on the prediction of real estate prices and trends in a certain urban area. Theoretical analysis and experimental results show that the proposed combined model can effectively solve the problem of insufficient extraction of data features and unsatisfactory prediction accuracy, and can achieve more accurate prediction effect than the single prediction model.