def get_search_space(target):
""" return search space for optimisation
target can be string, classifier, pipeline
"""
# get latest params
filename = os.path.dirname(__file__)
filename = os.path.join(filename, "classifiers.yaml")
with open(filename) as f:
allparams = yaml.load(f)
# classifier convert to string
if isinstance(target, BaseEstimator) \
and not isinstance(target, Pipeline):
target = get_clfname(target)
# string
if isinstance(target, str):
return allparams[target]
# pipeline
if isinstance(target, Pipeline):
# convert to list of (prefix, key) tuples
target = [(name, get_clfname(clf)) for name, clf in target.steps]
# process list of (prefix, key) into search space
search_space = {}
for prefix, key in target:
step_space = allparams[key]
step_space = {"{prefix}__{k}".format(prefix=prefix, k=k):v
for k,v in step_space.items()}
search_space.update(step_space)
return search_space
def crossval(self, verbose=0, seed=0, method="predict", **params):
""" returns crossval score
sets self.preds
"""
# track time spent per run
starttime = time()
np.random.seed(seed)
# useful for keras but throws exception for others
if "Keras" in get_clfname(self.clf):
self.clf.set_params(verbose=verbose)
self.clf.set_params(**params)
self.preds = cross_val_predict(self.clf, self.xtrain, self.ytrain,
method=method)
score = self.scorer._score_func(self.ytrain, self.preds) \
* self.scorer._sign
# log results
params.update(clf=get_clfname(self.clf),
name=self.name,
score=score,
elapsed=time()-starttime)
if self.runs:
self.runs.append(params, self.preds)
return score
def name(self):
""" label for charts and reporting """
if hasattr(self, "_name") and self._name is not None:
return self._name
if self.clf is not None:
return get_clfname(self.clf)
return None
def traintest(self, verbose=0, **params):
""" DEPRECATED. USE CROSSVAL.
return train/test score """
xtrain, xtest, ytrain, ytest = train_test_split(
self.xtrain, self.ytrain, test_size=0.1, random_state=0)
# verbose useful for keras. others throw exception.
if "Keras" in get_clfname(self.clf):
params["verbose"] = verbose
self.clf.fit(xtrain, ytrain, **params)
return self.scorer(self.clf, xtest, ytest)