Solar radiation forecasting is developed for electricity utilization in buildings to improve energy efficiency. In this paper, we first compare different machine learning methods for single-step solar radiation prediction, including multilayer perceptron (MLP), long short-term memory (LSTM), gradient boosting regression tree (GBRT), random forest (RF) and Informer. Then, we propose a hybrid model for short-term solar irradiation prediction, which is composed of a random forest model and an Informer model. Simulation results illustrate the effectiveness of our proposed hybrid model for balancing the accuracy-complexity tradeoff.