Prediction and Analysis of Chengdu Housing Rent Based on XGBoost Algorithm

Based on the online housing platform, this paper grabs 33224 pieces of data of Chengdu housing rental for visual analysis and prediction. Firstly, according to the importance of the characteristic attributes of the rental data, the characteristic factors such as area, housing area, orientation, rent collection mode, transportation, structure, etc. are selected. Through the data visualization method, it is found that Chengdu tenants have a greater demand for joint rental and small area apartment types. Then, 12 important features are predicted and fitted by RandomForestRegressor, XGBoost and LightGBM, and the best prediction accuracy is 0.85 on XGBoost model through parameter adjustment. Compared with RandomForestRegressor and LightGBM models, XGBoost model is more excellent in the application of this data set, and the prediction results have certain reference value for housing rental prices.