search

arundo / tsaug

A Python package for time series augmentation
https://tsaug.readthedocs.io
Apache License 2.0
347 stars 37 forks source link

Data augmentation for regression #7

Open mleeJordan opened 4 years ago

mleeJordan commented 4 years ago

First of all, many thanks to issue this open source library! I have been trying to model a time series regression with a limited time series data set. In particular, I have only 5000 data points (20 years of market/economic data) to model my companies returns. I have been searching how to augment time series and I came across this library. However, while checking the augmentation functions, I noticed that target has to be binary. So would it be possible to use this library for continuous target values?

Many thanks in advance!

Regards, Mehmet

tailaiw commented 4 years ago

I think it depends on the problem and the machine learning algorithm you will apply.

For example, if you want to warp the timeline for both X and Y, you may apply the time warping function with the same random seed to X and Y respectively, which will warp the timeline of X and Y in the same way.

For another example, adding noise to X and Y independently could be fine or problematic, depending on your problem and your downstream algorithm.

Unlike image augmentation methods (such as cropping, rotating) that are mostly universal, we found it is much trickier to select applicable augmentation methods for time series problems.