How is the value to split at chosen?

[ZmeiGorynych]

Hi, let's say you're considering to split a node along a certain float-valued dimension. How do you choose the candidate split values (that is, the value to which you compare the column values to decide if they end up in the right or left subtree)?
To choose among the candidates, you compare the error values - but how do you choose the candidates themselves?

[cerlymarco]

Hi, the candidate splits are chosen extracting quantiles from each feature distribution:

linear-tree/lineartree/_classes.py

Lines 336 to 341 in 2982edc

	# extract quantiles
	bins = np.linspace(0, 1, self.max_bins)[1:-1]
	bins = np.quantile(X, bins, axis=0)
	bins = list(bins.T)
	bins = [np.unique(X[:, c]) if c in self._categorical_features
	else np.unique(q) for c, q in enumerate(bins)]

If u support the project don't forget to leave a star ;-)
all the best

[ZmeiGorynych]

Perfect, thanks a lot!
I'm not using linear-tree directly right now, but am doing something very similar here, so just wanted to compare methodologies :)