Zhengzhou Second-hand Housing Price Evaluation Based On Machine Learning Model

In recent years,housing prices have experienced a process of increasing from year to year and gradually becoming more stable.Buyers not only pay attention to new houses,but also increasingly favor second-hand houses.In some cities,the second-hand housing market has developed rapidly,and the transaction volume has exceeded the volume of first-hand housing.With the prosperity of the second-hand housing market,the evaluation of second-hand housing prices will provide more intuitive market information for real estate developers,intermediaries and buyers,thus more effectively promoting second-hand housing transactions.In China the market method,income method and cost method are classic evaluation methods.They are applicable to different market environments.However,these methods are more focused on qualitative analysis and subject to subjective factors.With the development of computer technology,many scholars began to use computer software combined with machine learning methods to conduct a bulk evaluation of real estate prices.This method is more focused on quantitative analysis and more objective.This paper refers to the real estate appraisal method in recent years,using machine learning methods to evaluate the price of second-hand houses in Zhengzhou and achieved good results in empirical analysis.This article crawls the relevant data of second-hand housing in Zhengzhou City and analyzes the following three aspects:The first is descriptive statistical analysis.The relationship between the characteristic variables and the target variables is initially explored,and the results are visualized using Python software.The factors affecting housing prices were initially drawn,and the results were explained in light of the actual situation.The second is feature selection.By using two feature selection methods and combining the results of feature selection,the 18 features that are finally used for model training are obtained.The results of feature selection mainly involve factors such as the region,building area,building structure,number of households,number of elevators,and type of apartment,which are consistent with the reality.The third is to establish several models,respectively,to establish the Lasso regression model,the support vector machine model and the random forest model.In the modeling process,this paper uses the grid search method to adjust the parameters,and uses the mean square error as the evaluation criterion of the model,and compares the prediction effects of the three models on the test set.Finally,for the Zhengzhou second-hand housing data of this paper,the mean square error of the random forest is the smallest compared with the Lasso regression and support vector machine,and the scatter plot of the predicted value and the true value is more stably distributed near the straight line with the slope of 1,indicating Its prediction effect is more reliable.