The Prediction Of Rental Price And Analysis Of Its Influence Factors In Beijing

The housing price in Beijing is so high that it is beyond the financial capacity of most people.At the same time,the government strictly restricts the qualification of house purchase,which makes the struggling people in Beijing choose to rent a house to solve the housing problem.However,there is no unified rental price standard in the current rental market,which leads to the housing price gouging by real estate agents and seriously affects the healthy development of the rental market and people's normal life.Therefore,it is important to explore the influence factors of rent and how to use machine learning methods to predict rent in order to promote the healthy development of rental market.First of all,python and baidu map API are used to obtain the data of shared rental houses in Beijing in November 2018 from the website,the total data is 22042 pieces,including 18 characteristic variables,this provides data support for the following research.Secondly,histogram,boxplot,word cloud diagram,heat map and other statistical charts are used to carry out exploratory analysis and visualization on the influence factors of rent and spatial distribution of rent and the number of houses.It is concluded that rental price in Beijing is decreasing from the center to the outside,and has the characteristics of high in the north and low in the south,high in the east and low in the west.Some special location conditions lead to a small number of spatial variations.Thirdly,the random forest model and XGBoost model are established respectively by using training set data,and the parameters of the models are optimized,and the final prediction models of rental price are established.At the same time,the importance of variables is ranked.According to the ranking of the importance of variables,we can see that among the location characteristics,latitude and longitude,urban area,the number of subway stations and office buildings have a greater influence on the rental price,among the architectural features,the area,the type of house and whether independent toilet have a greater influence on the rental price.Finally,using the test set data and comparing the two final models through four common model prediction effect evaluation indexes,it can be seen that both the random forest model and the XGBoost model are reasonable in predicting the rental price,but XGBoost model has a better prediction effect in comparison.