def __init__(self, breaks, densities, bin_repr_points, scope=None, type_=None, meta_type=MetaType.DISCRETE):
Leaf.__init__(self, scope=scope)
self.type = type(self).type if not type_ else type_
self.meta_type = meta_type
self.breaks = breaks
self.densities = densities
self.bin_repr_points = bin_repr_points
def __init__(self, scope=None, data=None):
self._type = Type.BINARY
Leaf.__init__(self, scope=scope)
assert data is not None
self.n_features = data.shape[1]
def __init__(self, x_range, y_range, bin_repr_points, scope=None):
Leaf.__init__(self, scope=scope)
self._type = type
self.x_range = x_range
self.y_range = y_range
self.bin_repr_points = bin_repr_points
def __init__(self,
unique_vals,
mean,
inverted_mean,
square_mean,
inverted_square_mean,
prob_sum,
null_value_prob,
scope=None):
"""
Instead of histogram remember individual values.
:param unique_vals: all possible values in leaf
:param mean: mean of not null values
:param inverted_mean: inverted mean of not null values
:param square_mean: mean of squared not null values
:param inverted_square_mean: mean of 1/squared not null values
:param prob_sum: cumulative sum of probabilities
:param null_value_prob: proportion of null values in the leaf
:param scope:
"""
Leaf.__init__(self, scope=scope)
self.unique_vals = unique_vals
self.mean = mean
self.inverted_mean = inverted_mean
self.square_mean = square_mean
self.inverted_square_mean = inverted_square_mean
self.prob_sum = prob_sum
# ok, err = is_valid_prob_sum(self, self.unique_vals, n.cardinality)
# assert ok, err
self.null_value_prob = null_value_prob
def __init__(self,
breaks,
densities,
bin_repr_points,
scope=None,
meta_type=MetaType.REAL):
Leaf.__init__(self, scope=scope)
self.meta_type = meta_type
self.breaks = breaks
self.densities = densities
self.bin_repr_points = bin_repr_points
def __init__(self,
unique_vals,
probabilities,
null_value,
scope,
cardinality=0):
"""
Instead of histogram remember individual values.
:param unique_vals: all possible values in leaf
:param mean: mean of not null values
:param inverted_mean: inverted mean of not null values
:param square_mean: mean of squared not null values
:param inverted_square_mean: mean of 1/squared not null values
:param prob_sum: cumulative sum of probabilities
:param null_value_prob: proportion of null values in the leaf
:param scope:
"""
Leaf.__init__(self, scope=scope)
if not isinstance(unique_vals, np.ndarray):
unique_vals = np.array(unique_vals)
self.unique_vals = unique_vals
self.cardinality = cardinality
self.null_value = null_value
self.unique_vals_idx = None
self.update_unique_vals_idx()
# will be updated later
self.prob_sum = None
self.null_value_prob = None
self.mean = None
self.inverted_mean = None
self.square_mean = None
self.inverted_square_mean = None
if not isinstance(probabilities, np.ndarray):
probabilities = np.array(probabilities)
self.update_from_new_probabilities(probabilities)
def __init__(self, breaks, densities, bin_repr_points, scope=None):
Leaf.__init__(self, scope=scope)
self.breaks = breaks
self.densities = densities
self.bin_repr_points = bin_repr_points
def __init__(self, type, scope=None):
Leaf.__init__(self, scope=scope)
self._type = type
def __init__(self, vals, mean, scope=None):
Leaf.__init__(self, scope=scope)
self.vals = vals
self.mean = mean
def __init__(self, type, scope=None, weights=None, evidence_size=None):
Leaf.__init__(self, scope=scope)
self._type = type
self.evidence_size = evidence_size
self.weights = weights
def __init__(self, val, scope=None):
Leaf.__init__(self, scope=scope)
self.val = val