Unmanned aerial vehicles (UAVs), also known as drones, have many applications in various fields, and there has been a significant increase in their use in recent years. Many physical models of energy consumption in drones have been created. In recent years, data-driven models have emerged, achieving better results than models based on mathematical modelling of physical phenomena. This paper compares and analyses literature describing different models, providing background on factors affecting energy consumption and describing results achieved by different authors, including the most important physical models and focusing on machine learning models. The use of various models is discussed, including accurate results achieved by statistical models, such as XGBoost, decision trees and their modifications, including ensemble models. We show studies from this domain utilising artificial neural networks. We elaborate on the problems arising from using simple multilayer perceptron. We address the latest studies focused on advanced neural networks such as long short-term memory (LSTM) and temporal convolutional networks (TCN) that yield better results. Finally, further research is suggested, including gathering more data and using data augmentation, which may be beneficial in reducing overfitting in models, helping with model generalization, and improving their accuracy.