Based on the forecast data of the European Centre for Medium-Range Weather Forecasts (ECMWF) system and the observational data of the meteorological sites in Beijing, two machine learning algorithms (linear regression and gradient boosting regression tree) are used to correct the errors of apparent temperature. The prediction result was evaluated through the root mean square error (RMSE), and the result are also compared with the corrected results obtained by the traditional MOS method for further evaluation. The results show that the average RMSE predicted by linear regression, gradient-lifting regression tree, MOS and ECMWF model are 3.12, 3.06, 3.45 and 4.06℃, respectively. It is found that the results corrected through the machine learning algorithms are significantly better than those generated from MOS and Original forecast of ECMWF model. The machine learning correction method not only performs well in the plain area, but also achieve prominent results of correction effect with the data from high altitude sites, providing a certain technical guarantee for the normal operation of the Beijing Winter Olympics under complex mountain conditions.