Introduction

Time series is one of the fastest growing and richest types of data. In a variety of domains including dynamical systems, healthcare, climate science and economics, there have been increasing amounts of complex dynamic data due to a shift away from parsimonious, infrequent measurements to nearly continuous real-time monitoring and recording. This burgeoning amount of new data calls for novel theoretical and algorithmic tools and insights.

TorchTS is a deep learning library for time series forecasting built on Pytorch. The tools in this library mostly come out the research from Spatiotemporal Machine Learning Lab. The library is designed with minimal dependencies and user-friendly interfaces. A particular emphasis of the library is scalability, which exploits auto-differentiation and various inductive biases in the data.