def CalculatePi(self):
njobs = self.n_jobs
front_y = self.pareto_front_point()
front_y = self.y[front_y, :].T
predict_y_all = self.predict_y_all
del self.predict_y_all
gc.collect()
def tile_func(i, front_y0):
tile = 0
for front_y_i in front_y0.T:
big = i - front_y_i
big_bool = np.max(big, axis=1) < 0
tile |= big_bool
return tile
tile_all = parallelize(n_jobs=njobs,
func=tile_func,
iterable=predict_y_all,
front_y=front_y)
pi = np.sum(1 - np.array(tile_all), axis=1) / self.number
self.Pi = pi
return pi
def parallelize_score(self, inds):
"""
Parameters
----------
inds:list of SymbolTree
Returns
-------
list of (score,dim,dim_score)
"""
indss = [i.capsule for i in inds]
calls = functools.partial(calculate_collect, context=self.context, x=self.data_x, y=self.y,
terminals_and_constants_repr=self.terminals_and_constants_repr,
gro_ter_con=self.gro_ter_con, cv=self.cv, refit=self.refit,
dim_ter_con_list=self.dim_ter_con_list, dim_type=self.dim_type,
fuzzy=self.fuzzy,
scoring=self.scoring, score_pen=self.score_pen,
vector_add=self.vector_add,
add_coef=self.add_coef, inter_add=self.inter_add,
inner_add=self.inner_add, np_maps=self.np_map,
filter_warning=self.filter_warning,
dim_maps=self.dim_map, cal_dim=self.cal_dim)
score_dim_list = parallelize(func=calls, iterable=indss, n_jobs=self.n_jobs,
respective=False,
tq=self.tq, batch_size=self.batch_size)
return score_dim_list
def cal_binary_distance_all(self, slices=None, estimator_i=0):
""" calculate the distance matrix of slices """
self.estimator_i = estimator_i if isinstance(estimator_i,
int) else self.estimator_i
n_jobs = self.n_jobs
slices = slices if slices else self.slices
ret = self.check_prop("cal_binary_distance_all",
estimator_i=self.estimator_i,
slices=slices)
if ret is not None:
pass
else:
cal_binary_distance = partial(self.cal_binary_distance)
slices_cuple = list(itertools.product(slices, repeat=2))
ret = parallelize(n_jobs=n_jobs,
func=cal_binary_distance,
iterable=slices_cuple,
respective=True)
ret = np.reshape(ret, (len(slices), len(slices)), order='F')
self.add_prop("cal_binary_distance_all",
estimator_i=self.estimator_i,
slices=slices,
values=ret)
self.slices = slices
return ret
def cal_y_distance_all(self, slices=None, estimator_i=0):
""" calculate binary distance of 2 nodes """
self.estimator_i = estimator_i if isinstance(estimator_i,
int) else self.estimator_i
n_jobs = self.n_jobs
slices = slices if slices else self.slices
ret = self.check_prop("cal_y_distance_all",
estimator_i=self.estimator_i,
slices=slices)
if ret is not None:
pass
else:
cal_score = partial(self.cal_y_distance)
ret = parallelize(n_jobs=n_jobs, func=cal_score, iterable=slices)
self.add_prop("cal_y_distance_all",
estimator_i=self.estimator_i,
slices=slices,
values=ret)
self.slices = slices
return np.array(ret)
def _fit(self, x, y):
def score_pri(slices, x0, y0):
slices = list(slices)
if len(slices) < 1:
score0 = -np.inf
else:
slices = self.feature_unfold(slices)
data_x0 = x0[:, slices]
self.estimator.fit(data_x0, y0)
score0 = np.mean(self.estimator.best_score_) # score_test
# print(slices, score0)
return score0
score = partial(score_pri, x0=x, y0=y)
self.score_ = []
x, y = check_X_y(x, y, "csc")
assert all((self.check_must, self.check_muti)) in [True, False]
feature_list = list(range(x.shape[1]))
fold_feature_list = self.feature_fold(feature_list)
if self.check_must:
fold_feature_list = [
i for i in fold_feature_list if i not in self.check_must
]
slice_all = [combinations(fold_feature_list, i) for i in self.n_select]
slice_all = [
list(self.feature_must_fold(_)) for i in slice_all for _ in i
]
scores = parallelize(n_jobs=self.n_jobs,
func=score,
iterable=slice_all)
feature_combination = [self.feature_unfold(_) for _ in slice_all]
index = np.argmax(scores)
select_feature = feature_combination[index]
su = np.zeros(x.shape[1], dtype=np.bool)
su[select_feature] = 1
self.best_score_ = max(scores)
self.score_ = scores
self.support_ = su
self.estimator_ = clone(self.estimator)
if self.refit:
self.estimator_.fit(x[:, select_feature], y)
self.n_feature_ = len(select_feature)
self.score_ex = list(zip(feature_combination, scores))
self.scatter = list(zip([len(i) for i in slice_all], scores))
self.score_ex.sort(key=lambda _: _[1], reverse=True)
return self
def _fit(self, x, y, searchspace0, regclf0):
def fit_parllize(random_state):
data_train, y_train = sklearn.utils.resample(
x, y, n_samples=None, replace=True, random_state=random_state)
regclf0.fit(data_train, y_train)
predict_data = regclf0.predict(searchspace0)
predict_data.ravel()
return predict_data
njobs = self.n_jobs
number = self.number
predict_dataj = parallelize(n_jobs=njobs,
func=fit_parllize,
iterable=range(number))
return np.array(predict_dataj)
def cv_score_all(self, slices=None, estimator_i=0):
"""score all node with r2
Parameters
----------
slices : list, or None, default spath.slices
change to new slices to calculate
the lists of the index of feature subsets, each feature subset is a node,each int is the index of X
Examples 3 nodes
[[1,4,5],[1,4,6],[1,2,7]]
estimator_i: int, default spath.estimator_i
change to the estimator_i to calculate
Returns
----------
score_mean_std: nd.ndarray 2D
the mean and std
"""
self.estimator_i = estimator_i if isinstance(estimator_i,
int) else self.estimator_i
n_jobs = self.n_jobs
slices = slices if slices else self.slices
ret = self.check_prop("cv_score_all",
estimator_i=self.estimator_i,
slices=slices)
if ret is not None:
pass
else:
cal_score = partial(self.cv_score)
ret = parallelize(n_jobs=n_jobs, func=cal_score, iterable=slices)
self.add_prop("cv_score_all",
estimator_i=self.estimator_i,
slices=slices,
values=ret)
self.slices = slices
return np.array(ret)
def _cv_predict_all(self, slices=None, estimator_i=0):
""" calculate binary distance of 2 nodes """
self.estimator_i = estimator_i if isinstance(estimator_i, int) else self.estimator_i
n_jobs = self.n_jobs
batch_size = self.batch_size
slices = slices if slices else self.slices
ret = self.check_prop("cv_predict_all", estimator_i=self.estimator_i, slices=slices)
if ret is not None:
pass
else:
cal_score = partial(self.predict)
ret = parallelize(n_jobs=n_jobs, func=cal_score, iterable=slices, batch_size=batch_size)
self.add_prop("cv_predict_all", estimator_i=self.estimator_i, slices=slices, values=ret)
self.slices = slices
return ret
def fit(self, X, y, groups=None):
"""Fit the baf model and automatically tune the number of selected
feature.
Parameters
----------
X : {array-like, sparse matrix}, shape = [n_samples, n_feature]
Training vector, where `n_samples` is the number of samples and
`n_feature` is the total number of feature.
y : array-like, shape = [n_samples]
Target values (integers for classification, real numbers for
regression).
groups : array-like, shape = [n_samples], optional
cal_group labels for the samples used while splitting the dataset into
train/test set.
"""
X, y = check_X_y(X, y, "csr")
# Initialization
cv = check_cv(self.cv, y, is_classifier(self.estimator))
scorer = check_scoring(self.estimator, scoring=self.scoring)
ran = check_random_state(self.random_state)
baf = BackForward(
estimator=self.estimator,
n_type_feature_to_select=self.n_type_feature_to_select,
verbose=self.verbose,
primary_feature=self.primary_feature,
muti_grade=self.muti_grade,
muti_index=self.muti_index,
must_index=self.must_index,
random_state=ran)
func = partial(_baf_single_fit,
baf=baf,
estimator=self.estimator,
X=X,
y=y,
scorer=scorer,
random_state=ran)
scores = parallelize(n_jobs=self.n_jobs,
func=func,
iterable=cv.split(X, y, groups),
respective=True)
support, scores, score_step = zip(*scores)
best_support = support[np.argmax(scores)]
best_score = max(scores)
# Re-execute an elimination with best_k over the whole set
# Set final attributes
self.support_step = score_step
self.support_cv = support
self.support_ = best_support
self.score_cv = scores
self.score_ = best_score
self.estimator_ = clone(self.estimator)
self.estimator_.fit(X[:, self.support_], y)
self.n_feature_ = np.count_nonzero(support)
return self