The application of big data in the energy sector is considered as one of the main elements of Energy Internet. Crucial and promising challenges exist especially with the integration of renewable energy sources and smart grids. The ability to collect data and to properly use it for better decision-making is a key feature; in this work, the benefits and challenges of implementing big data analytics for renewable energy power stations are addressed. A framework was developed for the potential implementation of big data analytics for smart grids and renewable energy power utilities. A five-step approach is proposed for predicting the smart grid stability by using five different machine learning methods. Data from a decentralized smart grid data system consisting of 60,000 instances and 12 attributes was used to predict the stability of the system through three different machine learning methods. The results of fitting the penalized linear regression model show an accuracy of 96% for the model implemented using 70% of the data as a training set. Using the random forest tree model has shown 84% accuracy, and the decision tree model has shown 78% accuracy. Both the convolutional neural network model and the gradient boosted decision tree model yielded 87% for the classification model. The main limitation of this work is that the amount of data available in the dataset is considered relatively small for big data analytics; however the cloud computing and real-time event analysis provided was suitable for big data analytics framework. Future research should include bigger datasets with variety of renewable energy sources and demand across more countries.