Forecasting Time Series with VARMA Recursions on Graphs
Article in IEEE Transactions on Signal Processing by E. Isufi
Recent development in signal processing and network science has brought new tools for processing time series. However, it was not yet clear on how to exploit the structure of a network for predicting the evolution of time series. In this work, we leveraged tools from graph signal processing and autoregressive moving average recursions to do so. The findings show that by incorporating the graph structure into the model it is possible to improve the performance of standard vector autoregressive techniques and other state-of-the-art alternatives by a margin.
More can be found at https://ieeexplore.ieee.org/document/8767972