DeprecationWarning: scipy.signal.ricker is deprecated in SciPy 1.12 and will be removed

[Stanislav Khrapov]

Describe the bug

DeprecationWarning: scipy.signal.ricker is deprecated in SciPy 1.12 and will be removed
  in SciPy 1.15. We recommend using PyWavelets instead.
  
    wavelet_x = np.conj(ricker(points, width)[::-1])

To Reproduce

File functime_test.py

import polars as pl
from functime.feature_extractors import cwt_coefficients

cwt_coefficients(pl.Series([1 for _ in range(100)]))

In terminal:

pytest functime_test.py

Expected behavior

No DeprecationWarning

Desktop (please complete the following information):

• MacOSX 14.5
• Python 3.12.2
• functime==0.9.5

[baggiponte]

Thanks for the report! As far as I see, scipy 1.15 (and even 1.14) are still to be released. According to their milestones, scipy 1.15 should be due in 2025. Will address this with the upcoming releases.

I haven't decided the release cadence. I will blog soon about it but the idea is:

1. Keep coming the patches as they are needed
2. Prepare for a 0.10 stable release. functime is already doing forecasting and feature extraction pretty well. I need to make some minor breaking changes (such as the output format of conformal predictions) but aside that the core API won't change. This should be the release that welcomes polars 1.0.0 (there won't be a polars 0.21 series) and likely dropping py38.
3. 1.0.0 due well when we add all the features we'd like (e.g. bring-your-own-forecaster, neural net regressors...)

This will likely end up in the 0.10 release.

[baggiponte]

scipy 1.14 is released. I guess with v0.10 we can adapt this.