Multivariate time series prediction has wide applications in the area of financial investment, energy consumption, environmental pollution etc.
Because of the temporal complexity and nonlinearity existing in multivariate time series, few existing models could provide satisfactory prediction results
The traditional time series prediction models like Autoregressive Integrated Moving Average (ARIMA), widely used time series model, are not suitable for complex non linear multivariate time series.
Deep neural networks like RNNs, are now a days used to simulate the nonlinear structure to obtain better prediction results for multivariate time series . However, the serial calculation of RNNs slows down the training and prediction process with long sequences.
This repo is attempt made to develop the solution proposed in "Multivariate Time Series Prediction Based on Optimized Temporal Convolutional Networks with Stacked Auto-encoders" proposed by Wang et. al in Paper