What is it about?
We propose Radflow, a new forecasting model for networks of time series that influence each other. The model uses LSTMs and multi-head attention to learn the relationship between related time series. We achieve state-of-the-art results when forecasting traffic views on the network of YouTube videos and the network of Wikipedia articles.
Featured Image
Photo by Alina Grubnyak on Unsplash
Why is it important?
Our model Radlfow has the potential to improve forecasts in correlated time series networks such as the stock market, and impute missing measurements in ecology such as sensor data in networks of rivers.
Read the Original
This page is a summary of: Radflow: A Recurrent, Aggregated, and Decomposable Model for Networks of Time Series, April 2021, ACM (Association for Computing Machinery),
DOI: 10.1145/3442381.3449945.
You can read the full text:
Contributors
The following have contributed to this page
