Analysis And Forecast Of Housing Prices In Dalian City Based On Time Series Analysis

In recent years,the vigorous development of the real estate market has brought many new development opportunities for the real estate market in China.At the same time,the real estate industry is an important part and pillar industry of our national economy.How to grasp the trend of house price more accurately is of great significance and great influence to the development of real estate and the consumption decision and living standard of residents.This paper selects the monthly data of second-hand housing price in Ganjingzi District,Dalian City,in order to further improve the prediction accuracy,this paper uses the ARIMA model,exponential smoothing model,long and short-term memory network(Long Short-Term Memory,LSTM)model in time series analysis to analyze and predict the pre-processed data respectively,ARIMA model needs to carry out stationary test and white noise test before fitting,the purpose is to judge whether the sequence has certain rules to follow,rather than disorderly time series.This paper fits the appropriate ARIMA model after testing the sequence and obtains the prediction results for the next five periods;the advantage of the exponential smoothing model is that it can adjust its weight when predicting future values according to the time interval of historical data,which is more in line with the development law of most random events in reality;the LSTM network model has strong time series analysis ability,especially suitable for analyzing and predicting time series data with long-term dependence.A smaller root mean square error is obtained after prediction using LSTM network model.For analyzing the influence of many factors on house price,this paper selects several factors,such as the consumer price index of Dalian residents in manufacturing PMI?PPI?,the total population at the end of the year,and carries on the multivariate time series analysis combined with the house price data.It is found that the predicted values obtained by the time series model fitted in this paper are close to the actual values.the trend of ARIMA model fitting is more accurate;the prediction accuracy of exponential smoothing model is higher;LSTM network model has the highest prediction accuracy;the multiple time series analysis model has more accurate prediction trend but the error is the largest.Overall,house prices are still slowly rising in the coming months,with small short-term increases.