Research On Qingdao Housing Price Based On Additive Model

The real estate industry has an important position in our country's economy.At present,the methods of predicting the future trend of housing prices are widely used multi-factor regression models based on influencing factor analysis,single factor analysis models based on time series,and various housing price prediction models based on machine learning.Based on the analysis of the influencing factors of house prices,this paper considers the influence of non-parametric factors,and tries to predict the house prices in Qingdao based on the additive model.This paper first introduces the current research status of housing prices and additive models at home and abroad,and the development of Qingdao's real estate industry in the past 25 years,and analyzes the influencing factors of housing prices from both macro and micro aspects.Then it mainly introduces the relevant theoretical knowledge of additive model,including the form of additive model,estimation method and model checking method.Finally,this paper selects the relevant data of Qingdao from 1995 to 2019 to analyze and establish an additive partial linear model.When building the model,this article first conducts principal component analysis on influencing factors and selects the first three principal components for analysis.Then according to the scatter diagram,the first principal component is determined as the linear influence component,and the non-parametric influence of the second principal component and the third principal component on the housing price is statistically significant through single factor analysis.Therefore,this paper establishes an additive partial linear model(APLM)with the first principal component as the linear influencing factor,and the second and third principal components as the non-parametric influencing factors.Using R language to solve this model,we get that the variance explanation ability of this additive partial linear model is 99.7%,which means a good fit.Finally,by comparing with the fitting effect of partial linear model,it is found that the fitting effect of additive partial linear model is far better than that of partial linear model in the prediction of housing prices in Qingdao.