def _dataPreprocessing(self, X_l, X_u):
"""
Preprocess data: delete constant dimensions, Normalize input samples if needed
INPUT:
X_l Input data lower bounds (rows = objects, columns = features)
X_u Input data upper bounds (rows = objects, columns = features)
OUTPUT
X_l, X_u were preprocessed
"""
# delete constant dimensions
#X_l, X_u = delete_const_dims(X_l, X_u)
# Normalize input samples if needed
if X_l.min() < self.loLim or X_u.min() < self.loLim or X_u.max(
) > self.hiLim or X_l.max() > self.hiLim:
self.mins = X_l.min(axis=0) # get min value of each feature
self.maxs = X_u.max(axis=0) # get max value of each feature
X_l = normalize(X_l, [self.loLim, self.hiLim])
X_u = normalize(X_u, [self.loLim, self.hiLim])
else:
self.isNorm = False
self.mins = []
self.maxs = []
return (X_l, X_u)
def dataPreprocessing(self, X_l, X_u):
"""
Preprocess data: delete constant dimensions, Normalize input samples if needed
INPUT:
X_l Input data lower bounds (rows = objects, columns = features) datatype: numpy array
X_u Input data upper bounds (rows = objects, columns = features) datatype: numpy array
OUTPUT
X_l, X_u were preprocessed
"""
# delete constant dimensions
#X_l, X_u = delete_const_dims(X_l, X_u)
# Normalize input samples if needed
if X_l.min() < self.loLim or X_u.min() < self.loLim or X_u.max(
) > self.hiLim or X_l.max() > self.hiLim:
self.mins = torch.from_numpy(
X_l.min(axis=0)).float() # get min value of each feature
self.maxs = torch.from_numpy(
X_u.max(axis=0)).float() # get max value of each feature
X_l = normalize(X_l, [self.loLim, self.hiLim])
X_u = normalize(X_u, [self.loLim, self.hiLim])
else:
self.isNorm = False
self.mins = torch.FloatTensor([])
self.maxs = torch.FloatTensor([])
return (X_l, X_u)
def dataPreprocessing(self, Xh):
"""
Preprocess data: delete constant dimensions, Normalize input samples if needed
INPUT:
Xh Input data lower bounds (rows = objects, columns = features)
OUTPUT
Xh was preprocessed
"""
# delete constant dimensions
#Xh = delete_const_dims(Xh)
# Normalize input samples if needed
if Xh.min() < self.loLim or Xh.max() > self.hiLim:
self.mins = Xh.min(axis=0) # get min value of each feature
self.maxs = Xh.max(axis=0) # get max value of each feature
Xh = normalize(Xh, [self.loLim, self.hiLim])
else:
self.isNorm = False
self.mins = []
self.maxs = []
return Xh
