Here I am implementing some of the RNN structures, such as RNN, LSTM, and GRU to build an understanding of deep learning models for time-series forecasting. I have worked on some of the feature engineering techniques that are widely applied in time-series forecasting, such as one-hot encoding, lagging, and cyclical time features. The libraries used are Scikit-learn, Pandas, and PyTorch, an open-source machine learning library. I have followed this great article you can find more details over here link. It will give you extensive detail about this. Time Series Dataset I am using is from PJM’s Hourly Energy Consumption data, a univariate time-series dataset of 10+ years of hourly observations collected from different US regions. From all files, I am working with the PJM East region data, which originally has the hourly energy consumption data from 2001 to 2018, but you can use any dataset.
