scikit-learn FutureWarning

[sxjeru]

Describe the bug

.venv\Lib\site-packages\sklearn\utils\deprecation.py:151: FutureWarning: 'force_all_finite' was renamed to 'ensure_all_finite' in 1.6 and will be removed in 1.8. warnings.warn(

To Reproduce

import pmdarima as pm

model = pm.auto_arima(data, seasonal=False, trace=True)
print(model.summary())

Versions

python: 3.12.6 

Python dependencies:
 setuptools: 75.8.2
        pip: 25.0.1
    sklearn: 1.6.1
statsmodels: 0.14.4
      numpy: 1.26.4
      scipy: 1.15.2
     Cython: 3.0.12
     pandas: 2.2.3
     joblib: 1.4.2
   pmdarima: 2.0.4

Expected Behavior

No warning.

Actual Behavior

We got a lot of repeated warnings.

Additional Context

Follow the warning to change the name of the function used？

[jbbqqf]

Hi! Triaging older issues — this looks like a duplicate of the open PR #592.

The 'force_all_finite' was renamed to 'ensure_all_finite' in 1.6 and will be removed in 1.8 warning comes from sklearn 1.6's deprecation of force_all_finite in sklearn.utils.validation.check_array. pmdarima already has the right sentinel for the new name in pmdarima/utils/array.py:26-30:

_CHECK_ARRAY_FINITE_PARAM = (
    "ensure_all_finite"
    if "ensure_all_finite" in inspect.signature(skval.check_array).parameters
    else "force_all_finite"
)

…but a few call sites still pass force_all_finite=... as a literal kwarg (in pmdarima/arima/arima.py and pmdarima/utils/array.py). PR #592 ([MRG] Updated sklearn.utils.validation.check_array calls to use ensure_all_finite) makes exactly that fix. As long as #592 stays open the warning will keep showing up.

Would it make sense to close this in favour of #592 (and 👍 there to signal demand)?

(Disclosure: I drafted this comment with help from Claude Code while triaging stale issues; the cross-link to #592 was verified manually.)