Extend type hints in the classifiers and regressors

[adekusar-drl]

Extend type hints in the classifiers and regressors

[edoaltamura]

@adekusar-drl, could you please share an example or template of how you'd match the SKlearn type hinting for QML?

[adekusar-drl]

It has been a low priority for a long time, things might have changed. While scikit learn does not expose type hints by default, one of the possible options to take a look at https://numpy.org/doc/stable/reference/typing.html#module-numpy.typing. Numpy has ArrayLike type hint and it may work. This issue is more exploratory one rather something that can be easily implemented. Even if numpy's hinting is good enough, I'd first double check if such hinting brings in any value, not problems.

[edoaltamura]

For instance

qiskit-machine-learning/qiskit_machine_learning/algorithms/classifiers/pegasos_qsvc.py

Lines 128 to 130 in c59063a

	def fit(
	self, X: np.ndarray, y: np.ndarray, sample_weight: np.ndarray | None = None
	) -> "PegasosQSVC":

would become

import numpy.typing as npt

def fit( 
     self, X: npt.ArrayLike, y: npt.ArrayLike, sample_weight: npt.ArrayLike | None = None 
 ) -> "PegasosQSVC": 

or, more strictly,

import numpy.typing as npt

def fit( 
     self, X: npt.NDArray[np.float64], y: npt.NDArray[np.float64], sample_weight: npt.NDArray[np.float64] | None = None 
 ) -> "PegasosQSVC": 

While young, the new typing in Numpy is covered by tests and CI, and made it into a major release. So I'd consider the basic features relatively stable.