williamgilpin/dysts

混沌多元时间序列数据集，来源于混沌动力系统的多元时间序列。每个多元时间序列都是从混沌动力系统中提取的，适合用于长期预测任务。数据集包含400万个多元观测值，分为135个系统和三种粒度。子目录`coarse`、`medium`和`fine`各包含135个`.csv`文件，每个文件包含一个长度为10,000的多元时间序列。通道数量取决于具体的动力系统。由于混沌系统的遍历性，时间序列是平稳的。

The Chaotic Time Series Dataset is a multivariate time series dataset extracted from chaotic dynamical systems. The dataset contains 4 million multivariate observations, grouped into 135 systems and three granularities. Each subdirectory (coarse, medium, fine) contains 135 CSV files, each containing a single multivariate time series of length 10,000. The number of channels in the time series varies depending on the specific dynamical system, and the time series are stationary due to the ergodic property of chaotic systems. The dataset is suitable for long-horizon forecasting tasks.