关键词: 蝙蝠算法, BA-SVR&WSD混合模型, 网络搜索数据, 北京二手房预测

Abstract: Traditional SVR models can predict trends in house prices, but inappropriate parameter settings can affect the accuracy of predictions. In this paper, based on the nonlinear variation characteristics of Beijing second-hand house year-on-year price index, it introduces the Bat algorithm (BA) into the SVR model to optimize the three parameters of the model. Combing with web search data (Web Search Data, WSD), it builds the BA-SVR&WSD mixed model, and gives the prediction process of the model algorithm. It also carries out the comparison study by introducing multiple benchmark prediction models and predictive performance metrics. The research results show that the SVR model based on bat algorithm has better generalization ability,more accurate prediction effect and higher prediction accuracy. This prediction method also provides a valuable reference for the monitoring and regulation of second-hand houseinpricesin Beijing.

Key words: Bat Algorithm, BA-SVR&WSDMixedModel, Web Search Data, Beijing Second-hand Housing Forecast