Research On The Prediction Of Sales Time For New Urban Commercial Housing Based On Neural Networks

The real estate industry is one of the pillar industries in China’s national economy,which affects the overall macroeconomic stability of China.All long,the problem of a long period of housing destocking cycle has always been affecting the sustainable development of the real estate industry,particularly in third and fourth tier cities.Some real estate enterprises have been affected by this problem,resulting in problems in fund circulation.In order to solve this problem,this article takes the depreciation time of newly built commercial housing in Handan City as the research object,and based on the current national policies,under the premise of stable economic environment and no force majeure effects,conducts a correlation prediction study between housing prices and housing destocking time.Firstly,as a commodity,the fluctuation in house prices itself will affect its destocking time.To determine the benchmark point of the pricing fluctuation interval for the house,the monthly average prices of commodity housing in Handan City were analyzed,and a moving autoregressive integrated moving average model(ARIMA)and a long short-term memory neural network(LSTM)model were established,as well as an ARIMA-LSTM hybrid model for prediction.The evaluation indicators,such as mean absolute error(MAE),root mean square error(RMSE),and coefficient of determination(R2),were used to evaluate the prediction results of the model.The results showed that the proposed hybrid model had the lowest error and the best performance compared to the other two models.Select the optimal prediction model,whose prediction results serve as the benchmark for predicting the price fluctuation range of the housing destocking time.Then,a BP neural network model based on clustering prediction,,a BP neural network model optimized by genetic algorithm(GA-BP),and a LSTM model based on sequence prediction were used separately to establish the predictive model for housing destocking time.The evaluation indicators,such as MAE,RMSE,and R~2,were used to evaluate the accuracy of the models.The results showed that the housing destocking time can be predicted by establishing a model based on related influencing factors.From the perspective of evaluation indicators,the predictive effect of the LSTM model which based on sequence prediction is better than the other two algorithm models which based on cluster prediction.Finally,based on the predicted monthly average house price as the benchmark point,the fluctuation interval of the house price was determined according to the historical fluctuation level of the house price,and the destocking time of the designated house was predicted,and the correlation curve between house price and destocking time.By observing and analyzing the curve,the conclusion was drawn that,overall,there is a positive correlation between house price and destocking time,while in a small range of house price fluctuations,there is no stable positive correlation between house price and destocking time.This study established a predictive model for monthly average house price and house destocking time,and the results showed that house turnover time can be predicted by adjusting the house price,and the house price-destocking time prediction curve can be obtained,which provides a reference for real estate enterprises to formulate marketing decisions.