The purpose of this study was to apply the random forest (RF), XGBoost, and LightGBM machine learning (ML) algorithms to land cover classification, and to present the model tuning process for each algorithm. Sentinel-2 satellite images were used for land cover classification, and the land cover map provided by the Ministry of Environment of the Republic of Korea was used as label data. Each ML algorithm was applied using the constructed dataset. In addition, each ML algorithm was optimized by three methods (grid search, random search, and Bayesian optimization). The grid search took the longest time to optimize the hyperparameters because it required the highest number of search iterations, but the accuracy was highest. The random search was the fastest method of optimizing the hyperparameters. The accuracy of XGBoost was the highest for each ML algorithm. The prediction of XGBoost was the most consistent with the land cover map provided by the Ministry of Environment. However, the LightGBM algorithm has a major advantage in terms of the algorithm optimization and application time. Therefore, our study is meaningful in that we obtained a higher accuracy and shorter time for each ML algorithm.