Shulou(Shulou.com)06/02 Report--

This article to share with you is about Python in the probability time series model is what, Xiaobian think quite practical, so share to everyone to learn, I hope you can read this article after some gains, not much to say, follow Xiaobian to see it. Title:

GluonTS: Probabilistic Time Series Models in Python

Author:

Alexander Alexandrov, Konstantinos Benidis, Michael Bohlke-Schneider, Valentin Flunkert, Jan Gasthaus, Tim Januschowski, Danielle C. Maddix, Syama Rangapuram, David Salinas, Jasper Schulz, Lorenzo Stella, Ali Caner Türkmen, Yuyang Wang

Source:

Machine Learning (cs.LG)

Submitted on 12 Jun 2019 (v1), last revised 14 Jun 2019

Document link:

arXiv:1906.05264

Code Link:

https://github.com/awslabs/gluon-ts

abstract

We'll introduce Gluon Time Series (GluonTS, available at this https URL), a library for deep learning-based time series modeling. Gluons simplify the development and experimentation of time series models for common tasks such as prediction or anomaly detection. It provides scientists with all the necessary components and tools to quickly build new models, run and analyze experiments efficiently, and assess model accuracy.

original English text

We introduce Gluon Time Series (GluonTS, available at this https URL), a library for deep-learning-based time series modeling. GluonTS simplifies the development of and experimentation with time series models for common tasks such as forecasting or anomaly detection. It provides all necessary components and tools that scientists need for quickly building new models, for efficiently running and analyzing experiments and for evaluating model accuracy.

points

The structure of this article is as follows. In Section 2, we discuss the general design principles and architecture of the library and discuss the different components available in GluonTS. In Section 3, we formally introduced a series of time series problems that gluons allow to solve. Section 4 outlines common neural prediction architectures that can be assembled with gluons or implemented as pre-bound baseline models. In Section 5, we ran benchmark models of published deep learning-based predictive models based on 11 public datasets and demonstrated the applicability of these models to other tasks such as anomaly detection. We discuss related work in Section 6 and summarize future work in Section 7.

We introduced GluonTS, a toolkit for building time series models based on deep learning and probabilistic modeling techniques. By providing tools and abstractions such as probabilistic models, basic neural building blocks, human-readable model logs to improve repeatability and uniform I/O & gluon evaluation, scientists are able to rapidly develop new time series models for common tasks such as prediction or anomaly detection. GluonTS has been actively used in a variety of internal and external use cases at Amazon, including production, helping scientists solve time series modeling challenges.

GluonTS 'pre-bound state-of-the-art model implementation allows simple benchmarking of new algorithms. We demonstrated this in large-scale experiments running the pre-bound model on different datasets and compared its accuracy to classical methods. Such experiments are the first step towards a deeper understanding of the neural architecture of time series modeling. More fine-grained experiments, such as ablation experiments and controlled data experiments, are needed next. Stickers provide the necessary tools for future work.

Table: Comparison of results between models on different data

The above is what the probabilistic time series model in Python is. Xiaobian believes that some knowledge points may be what we see or use in our daily work. I hope you can learn more from this article. For more details, please follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.