Skip to content

Add support for ConfigSpace.hyperparameters.Constant #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
nchristensen wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Add support for ConfigSpace.hyperparameters.Constant #3

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
41fdded
Add support for ConfigSpace.hyperparameters.Constant
nchristensen Dec 11, 2022
251b609
Run autopep8 on space.py
nchristensen Dec 11, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
71 changes: 49 additions & 22 deletions skopt/space/space.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,4 @@
from sklearn.impute import SimpleImputer
import numbers
import numpy as np
import yaml
Expand All @@ -10,7 +11,7 @@
from sklearn.utils.fixes import sp_version


if type(sp_version) is not tuple: # Version object since sklearn>=2.3.x
if not isinstance(sp_version, tuple): # Version object since sklearn>=2.3.x
if hasattr(sp_version, "release"):
sp_version = sp_version.release
else:
Expand All @@ -34,9 +35,10 @@
ccs_active = True
except (ImportError, OSError) as a:
import warnings
warnings.warn("CCS could not be loaded and is deactivated: " + str(a), category=ImportWarning)

from sklearn.impute import SimpleImputer
warnings.warn(
"CCS could not be loaded and is deactivated: " +
str(a),
category=ImportWarning)


# helper class to be able to print [1, ..., 4] instead of [1, '...', 4]
Expand Down Expand Up @@ -121,7 +123,8 @@ def check_dimension(dimension, transform=None):

if len(dimension) == 2:
if any(
[isinstance(d, (str, bool)) or isinstance(d, np.bool_) for d in dimension]
[isinstance(d, (str, bool)) or isinstance(d, np.bool_)
for d in dimension]
):
return Categorical(dimension, transform=transform)
elif all([isinstance(dim, numbers.Integral) for dim in dimension]):
Expand Down Expand Up @@ -242,11 +245,13 @@ def _uniform_inclusive(loc=0.0, scale=1.0):
# XXX scale is very large.
return uniform(loc=loc, scale=np.nextafter(scale, scale + 1.0))


def _normal_inclusive(loc=0.0, scale=1.0, lower=-2, upper=2):
assert lower <= upper
a, b = (lower - loc) / scale, (upper - loc) / scale
return truncnorm(a, b, loc=loc, scale=scale)


class Real(Dimension):
"""Search space dimension that can take on any real value.

Expand Down Expand Up @@ -287,6 +292,7 @@ class Real(Dimension):
can be float.

"""

def __init__(self, low, high, prior="uniform", base=10, transform=None,
name=None, dtype=float, loc=None, scale=None):
if high <= low:
Expand Down Expand Up @@ -347,7 +353,8 @@ def set_transformer(self, transform="identity"):
self._rvs = _uniform_inclusive(0.0, 1.0)
assert self.prior in ["uniform", "log-uniform"]
if self.prior == "uniform":
self.transformer = Pipeline([Identity(), Normalize(self.low, self.high)])
self.transformer = Pipeline(
[Identity(), Normalize(self.low, self.high)])
else:
self.transformer = Pipeline(
[
Expand All @@ -363,7 +370,8 @@ def set_transformer(self, transform="identity"):
self._rvs = _uniform_inclusive(self.low, self.high - self.low)
self.transformer = Identity()
elif self.prior == "normal":
self._rvs = _normal_inclusive(self.loc, self.scale, self.low, self.high)
self._rvs = _normal_inclusive(
self.loc, self.scale, self.low, self.high)
self.transformer = Identity()
else:
self._rvs = _uniform_inclusive(
Expand All @@ -375,7 +383,7 @@ def set_transformer(self, transform="identity"):

def __eq__(self, other):
return (
type(self) is type(other)
isinstance(self, type(other))
and np.allclose([self.low], [other.low])
and np.allclose([self.high], [other.high])
and self.prior == other.prior
Expand Down Expand Up @@ -607,7 +615,8 @@ def set_transformer(self, transform="identity"):
self._rvs = randint(self.low, self.high + 1)
self.transformer = Identity()
elif self.prior == "normal":
self._rvs = _normal_inclusive(self.loc, self.scale, self.low, self.high)
self._rvs = _normal_inclusive(
self.loc, self.scale, self.low, self.high)
self.transformer = ToInteger()
else:
self._rvs = _uniform_inclusive(
Expand All @@ -619,7 +628,7 @@ def set_transformer(self, transform="identity"):

def __eq__(self, other):
return (
type(self) is type(other)
isinstance(self, type(other))
and np.allclose([self.low], [other.low])
and np.allclose([self.high], [other.high])
)
Expand Down Expand Up @@ -649,7 +658,8 @@ def inverse_transform(self, Xt):
if self.dtype == int or self.dtype == "int":
# necessary, otherwise the type is converted to a numpy type
return getattr(
np.round(inv_transform).astype(self.dtype), "tolist", lambda: value
np.round(inv_transform).astype(
self.dtype), "tolist", lambda: value
)()
else:
return np.round(inv_transform).astype(self.dtype)
Expand Down Expand Up @@ -740,7 +750,8 @@ def __init__(self, categories, prior=None, transform=None, name=None):
self.prior = prior

if prior is None:
self.prior_ = np.tile(1.0 / len(self.categories), len(self.categories))
self.prior_ = np.tile(1.0 /
len(self.categories), len(self.categories))
else:
self.prior_ = prior
self.set_transformer(transform)
Expand All @@ -755,7 +766,8 @@ def set_transformer(self, transform="onehot"):

"""
self.transform_ = transform
if transform not in ["identity", "onehot", "string", "normalize", "label"]:
if transform not in ["identity", "onehot",
"string", "normalize", "label"]:
raise ValueError(
"Expected transform to be 'identity', 'string',"
"'label' or 'onehot' got {}".format(transform)
Expand All @@ -780,11 +792,12 @@ def set_transformer(self, transform="onehot"):
self._rvs = _uniform_inclusive(0.0, 1.0)
else:
# XXX check that sum(prior) == 1
self._rvs = rv_discrete(values=(range(len(self.categories)), self.prior_))
self._rvs = rv_discrete(
values=(range(len(self.categories)), self.prior_))

def __eq__(self, other):
return (
type(self) is type(other)
isinstance(self, type(other))
and self.categories == other.categories
and np.allclose(self.prior_, other.prior_)
)
Expand Down Expand Up @@ -926,7 +939,8 @@ def __init__(self, dimensions):
vals = list(x.choices)
if x.name in cond_hps:
vals.append("NA")
param = Categorical(vals, prior=x.probabilities, name=x.name)
param = Categorical(
vals, prior=x.probabilities, name=x.name)
space.append(param)
self.hps_type[x.name] = "Categorical"
elif isinstance(x, CS.hyperparameters.OrdinalHyperparameter):
Expand All @@ -936,6 +950,13 @@ def __init__(self, dimensions):
param = Categorical(vals, name=x.name)
space.append(param)
self.hps_type[x.name] = "Categorical"
elif isinstance(x, CS.hyperparameters.Constant):
vals = [x.value]
if x.name in cond_hps:
vals.append("NA")
param = Categorical(vals, name=x.name)
space.append(param)
self.hps_type[x.name] = "Categorical"
elif isinstance(x, CS.hyperparameters.UniformIntegerHyperparameter):
prior = "uniform"
if x.log:
Expand All @@ -962,7 +983,8 @@ def __init__(self, dimensions):
elif isinstance(x, CS.hyperparameters.NormalFloatHyperparameter):
prior = "normal"
if x.log:
raise ValueError("Unsupported 'log' transformation for NormalFloatHyperparameter.")
raise ValueError(
"Unsupported 'log' transformation for NormalFloatHyperparameter.")
param = Real(x.lower, x.upper, prior=prior, name=x.name,
loc=x.mu, scale=x.sigma)
space.append(param)
Expand All @@ -989,7 +1011,8 @@ def __init__(self, dimensions):
if x.name in cond_hps:
vals.append("NA")
if isinstance(distrib, CCS.RouletteDistribution):
param = Categorical(vals, prior=distrib.areas, name=x.name)
param = Categorical(
vals, prior=distrib.areas, name=x.name)
elif isinstance(distrib, CCS.UniformDistribution):
param = Categorical(vals, name=x.name)
else:
Expand All @@ -1007,7 +1030,8 @@ def __init__(self, dimensions):
elif isinstance(distrib, CCS.NormalDistribution):
prior = "normal"
if distrib.scale_type == CCS.ccs_scale_type.LOGARITHMIC:
raise ValueError("Unsupported 'log' transformation for CCS.NumericalHyperparameter with normal prior.")
raise ValueError(
"Unsupported 'log' transformation for CCS.NumericalHyperparameter with normal prior.")
else:
raise ValueError("Unsupported distribution")
if CCS.ccs_numeric_type.NUM_INTEGER:
Expand Down Expand Up @@ -1092,7 +1116,10 @@ def from_yaml(cls, yml_path, namespace=None):
with open(yml_path, "rb") as f:
config = yaml.safe_load(f)

dimension_classes = {"real": Real, "integer": Integer, "categorical": Categorical}
dimension_classes = {
"real": Real,
"integer": Integer,
"categorical": Categorical}

# Extract space options for configuration file
if isinstance(config, dict):
Expand Down Expand Up @@ -1132,7 +1159,6 @@ def _cs_post_process_conf(self, hps_names, conf):
point.append(val)
return point


def _ccs_post_process_conf(self, hps_names, conf):
point = []
values = conf.values
Expand Down Expand Up @@ -1311,7 +1337,8 @@ def inverse_transform(self, Xt):
if offset == 1:
columns.append(dim.inverse_transform(Xt[:, start]))
else:
columns.append(dim.inverse_transform(Xt[:, start : start + offset]))
columns.append(dim.inverse_transform(
Xt[:, start: start + offset]))

start += offset

Expand Down