Forecast Of Second-hand Housing Transaction Price In Beijing Based On GA-BP Model

At present,the types of residential buildings in China mainly include commercial housing and affordable housing.Commercial housing has risks such as imperfect surrounding facilities and invisible building quality.However,there are limitations in the application of housing qualifications and limited transactions.This article turned the research object to the second-hand housing market.The research on second-hand housing prices mainly focuses on the average housing price and the listing price of houses.However,the research on the average price of second-hand housing has no practical guiding significance for the purchasers of houses,and the listing price of second-hand houses is usually higher than that of houses.The actual value,the conclusion of its research will also lead to wrong guidance for consumers.Therefore,the research on the transaction price forecast of second-hand houses is of great practical significance.This paper takes the transaction price of second-hand housing in Beijing as the research object,and uses the online housing sales organization as the data source.From January to March 2019,a total of 9196 Beijing second-hand housing transaction data were selected.When selecting relevant indicators that affect the transaction price of second-hand houses in Beijing,this paper conducts research on four aspects: location characteristics,architectural features,neighborhood environment,and transaction indicators,and finally selects 38 impact indicators.In order to ensure the prediction effect of the model,the amount of valid data finally determined by data preprocessing means is 8985.This paper predicts the transaction price of second-hand housing in Beijing by constructing GA-BP model.The results show that with the increase of the number of iterations of genetic algorithm,the training error of GA-BP model shows a downward trend and converges to 27 times.The optimal solution,the training error at this time is 0.0077,which means that the weight and threshold of the BP neural network have been optimized.In the final forecasting stage,the GA-BP model has a prediction error of 0.012,indicating that the GA-BP model accurately predicts the transaction price of second-hand houses in Beijing.