Energy forecasting and planning play an important role in energy sector development and policy formulation. The forecasting model selection mostly based on the availability of the data, and the principal objective is the planning exercise and tool in different modern energy systems. Existing literature explicate that the supervised based machine learning algorithms are intelligent for predictions such as future load demand, wind, solar and geothermal energy forecasting, etc., across the wide range of applications. In this study, a comprehensive review is conducted on supervised based machine learning algorithms by using three well-known forecasting engines. This review aims to suggest suitable methods for forecasting analysis and several other prediciton tasks. A particular objective is to investigate and analyzed the methods used to forecast energy, real time use in multiple applications and to identify the research review with useful techniques that are approachable in the current literature. This review contains a critical comparison and study among various modeling techniques to choose a better forecasting model for performing the desired task in multiple applications. The forecasting accuracy is compared and analyzed through comprehensive literature review and with the real-time energy consumption and climate data used for modeling analysis. The Bayesian regularization backpropagation neural networks (BRBNNs) and the Levenberg Marquardt backpropagation neural networks (LMBNNs) render higher forecasting accuracy and performance with the coefficient of correlation 0.972 and 0.971 respectively. The supervised learning is useful for real-time applications because of optimal scenario of a specific model will allow for the model to precisely predict the label's class for unseen data instances. Additionally, the supervised learning can reduce the noise and control imbalanced data in the network formulation.