A Study on Housing Price Forecasts Based on Web Search Indices

With the development of the Internet, the web search index has become an important data source for economic and market forecasting. This paper investigates the prediction of new residential price index in Nanjing based on web search index. Keywords are extracted from news text, keyword expansion and screening are performed, and Baidu index is used to obtain search data, which is modeled by combining differential autoregressive moving average model (ARIMA), random forest, and long-short-term memory neural network (LSTM) model. The experimental results show that the web search data are significantly associated with the house prices in Nanjing, and the prediction model constructed using the search index performs well, with the best fitting effect of the random forest model. This study proves the feasibility of web search index in house price prediction and provides a new perspective for market analysis.