This is my team's second time series related library that we've recently open sourced, the other being https://github.com/arundo/adtk, our anomaly detection toolkit for time series.

We created this library because we were training a lot of deep learning models on time series data, but needed more examples of the specific types of behaviour we were interested in. This data augmentation library is inspired by image data augmentation libraries, but taking the considerations of time series into account. As with image data, you need to carefully consider whether the specific "augmentation" will preserve the aspects of the data that you are interested in.

We've released tsaug under an Apache license. We'd love to have people try it out, make contributions, and ask any questions.

tsaug is pip installable and the documentation and examples are linked in the readme on Github.

Credit primarily goes to Tailai Wen, who led this effort.

Here is a blog article I wrote to introduce the package https://www.arundo.com/arundo_tech_blog/tsaug-an-open-source...