The project is in HKU deep learning lab. We analyze a large set of energy consumption data and build real time application based on the prediction. The topic involves different machine learning approaches and comparison. Since the machine learning is often discussed, I will also talk about the often ignored skill in data science. How to handle huge data and optimize the processing code, how to effectively build useful tools and make the life of data scientist easier, and how to correctly interpret the predicting result to enhance the project level from merely "machine learning" to real "data-driven application"