Be careful with np.random.seed

[mpvanderschelling]

The problem

What can happen, is that you initialize this class with a seed, an set the numpy RNG: (numpy.random.set_seed(seed))
Then in your code, you decide to set the seed again, but then the np.random.seed is changed and the class seed is also changed unintenionally.

Possible solution

Use numpy generator, this is advised by the numpy community.

Assignees:
Labels:
Milestone:

mfbml/src/mfbml/methods/kernel_ridge_regression.py

Lines 68 to 70 in 1038d86

	# set random seed
	self.seed = seed
	np.random.seed(seed)