The design of sustainable spaces has become of great importance due to the impact of thermal comfort in urban open spaces on the feeling of comfort and satisfaction of citizens, as well as the optimal use of environmental conditions in creating comfort and saving energy consumption. Accurate prediction of thermal comfort and environmental thermal conditions in urban areas by various methods facilitates the improvement of urban planning and energy management. A review of previous studies shows the high ability of deep learning models in improving thermal comfort predictions. In this paper, a hybrid method based on deep learning is presented for predicting the universal thermal climate index. After preprocessing the dataset, the previous models that had high accuracy were trained and evaluated. Then, the three models with the highest performance were selected for the combined model, and the predictions of all three models were generated for the test set. To ensure dimensionality integrity, the predictions were converted into one-dimensional arrays. Finally, the final prediction was calculated by averaging the predictions of the three models. To evaluate the performance of the proposed model and to train and test the methods used in the combined model, the real data set of Mashhad city was used. Standard numerical prediction criteria including root mean square error, mean absolute error and coefficient of determination were calculated to evaluate the proposed model. The evaluation results showed that the hybrid model provides better performance in predicting the universal thermal climate index compared to previous methods.