search

functime-org / functime

Time-series machine learning at scale. Built with Polars for embarrassingly parallel feature extraction and forecasts on panel data.
https://docs.functime.ai
Apache License 2.0
970 stars 52 forks source link

Fix test bug in test_diff #216

Closed abstractqqq closed 1 month ago

abstractqqq commented 1 month ago

Reference Issues/PRs

What does this implement/fix? Explain your changes.

Fixed the test_diff in test_preprocessing.py

Any other comments?

The test failed because of float point numerical issue when re-constructing original X. The easiest way is to not re-construct the data by applying inverse transform. Just copy the data to represent the original. The downside of this fix is that the .invert method is not tested. Real world time series practitioners can argue about the merit of having a .invert method on the transform. To me personally, I don't think it matters (especially when memory is abundant)

vercel[bot] commented 1 month ago

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name Status Preview Comments Updated (UTC)
functime-docs ✅ Ready (Inspect) Visit Preview 💬 Add feedback May 30, 2024 0:43am
baggiponte commented 1 month ago

Ciao TQ, thank you very much for the contribution. Will review later 😊