Classification Research Of Resold House Data Based On Ensemble Learning

The real estate shows more and more significance with the development of economy in our country.Although after more than ten years of rapid growth in houses’ price,the popularity of real estate is a little lower compared with before,it still attracts abundant attention,especially in the field of resold houses.This thesis constructs a effective classification model of high quality resold houses by ensemble learning.This classification model can provide constructive opinions for ordinary people who wants to purchase house,estate developers when they develop real estates and the intermediary to promote houses.This thesis uses multiple data preprocessing methods to preprocess the data of resold houses which can be used directly in the analyzing process later,such as data cleaning,missing value filling,data standardization and data dimension reduction.This thesis designs and constructs three classification models of high quality resold houses based on ensemble learning.The first model is a random forest model integrated by decision tree.Through the method of adaboost algorithm,this thesis also uses a neural network ensemble model constructed by multi-layer perception neural networks which is the second one.Besides,the last model is an improved deep forest model.After constructing three resold housing models,this thesis analyzes and compares these three models through comparative experiments.The thesis uses several evaluation standards for the three ensemble classification models in classifying resold houses data,and the result shows that all of the three models can achieve wonderful results in classification scenes.