基于BP神经网络对房价预测的研究

Prediction of House Price Based on BP Neural Network

搜集西安市1997年~2016年20年七个对房价影响的数据,首先对数据进行预处理,包括数据集成、空白值填写、数据标准化,然后,使用PCA方法对数据进行降维处理,在干净数据的基础上,采用对数函数logsig为激励函数,学习率为0. 3,建立BP神经网络的房价预测模型,选择80%的数据作为训练数据,20%的数据作为测试数据。通过MATLAB对数据进行仿真实验,其实验结果表明用该模型仿真的结果其正确率较高,最高达到94. 14%,效果较好,具有一定的实用价值。

From 1997 to 2016,collecting data on the impact of Xi＇an city on house prices, and the data were pretreated firstly ,including data integration,blank value filling and data standardization. Then ,the data were reduced by PCA method. On the basis of clean data, the logarithmic function logsig was used as an incentive function, and the learning rate was 0. 3. The price forecasting model of BP neural network was established. 80% of the data were selected as training data and 20% of the data were used as testing data. The simulation experiment of the data was cmTied out through MAT- LAB. The experimental resuhs showed that the simulation resuh of the model is high,up to 94. 14% ,the effect is better, and has certain practical value