Prediction Of Kunming Commodity Price Index Based On Keyword Search

With the rapid development of China's economy,the real estate industry has been rising rapidly and become the pillar industry of the national economy.Especially in the last decade,the continue increasing of real estate price has attracted the focus of whole society.The ultra-high commercial housing price has a great impact on the operation of the national macro-economy and the quality of life of the residents.Therefore,the accurate grasp of the price index of commercial housing is essential to make scientific and reasonable decisions at the national and local levels.It is noted that the traditional commercial housing price index data mostly comes from the National Bureau of statistics system,and the survey data are seriously lagged behind,which will affect the efficiency of prediction.Nowadays,people tend to rely on the internet for various kind of information.In the big data era,the digging and analysis of data from the internet could be used to precisely predict the market demand and the trend of consumer's behavior.This paper uses Python to crawl 68 key words related to the Kunming commercial housing price index from the Baidu index,10 of these key words are finally selected based on the correlation coefficient and the random forest algorithm.These 10 key words are: the ultimate real estate network,large-sized apartment renovation renderings,mortgage interest rates,China building materials market,building materials,China's housing provident fund loans,housing mortgage the loan interest rate,rental contract,rental kunming and kunming soufun.Then,the 10 keywords are used as predictive variables.The genetic algorithm and particle swarm optimization algorithm were respectively used to optimize the parameters in the least squares support vector machine regression.The results showed that the parameter error obtained by the particle swarm optimization algorithm was slightly smaller.Therefore,the solution is based on the least squares optimization particle swarm optimization algorithm and the result is used for the prediction of Kunming commercial housing price index.The result of prediction is significant,the goodness of fit is 0.9441,the average absolute error is 0.6617 and the mean square error is 0.8073.The results show that consumers' search behavior is a reflection of real estate transactions,and to a large extent,can affect consumers' decisions and behavior.Besides,the method of forecasting house price by keywords canprovide some references for government in making relevant policies and provide better guidance for people in relevant sectors.