Adaptive Learning for Time Series Forecasting

There is no need to say the importance of time series forecasting applications in various industries from Energy to Healthcare, etc. Therefore, let’s go to the point directly. One of the complex and difficult challenges that we can face while working on time series datasets is their variety in statistical features, which can lead to shifts in their distributions and, consequently, various behaviors that make them difficult to understand by models. This article provides a two-stage model to deal with Temporal Covariate Shift (TCS); we call it ADaRNN (combination of Adaptive…