def distance(self, X, Y, **kwargs):
""" Step 1: project X and Y in the learned latent space,
Step 2: compute distance between the projections (NNAA score by default).
"""
X_enc = self.encoder.predict(X)
Y_enc = self.encoder.predict(Y)
if not isinstance(X_enc, autopandas.AutoData):
X_enc = autopandas.AutoData(X_enc)
if not isinstance(Y_enc, autopandas.AutoData):
Y_enc = autopandas.AutoData(Y_enc)
return X_enc.distance(Y_enc, **kwargs)
def copula_generate(X, generator=None, n=None):
""" Generate using copula trick.
:param generator: Model to fit and sample from. KDE by default.
:param n: Number of examples to generate. By default it is the number of observations in X.
"""
indexes = X.indexes
columns = X.columns
if generator is None:
generator = KernelDensity()
if n is None:
n = X.shape[0]
X_real = np.array(X)
# X marginals to uniforms
X = matrix_to_rank(X)
# X uniforms to inverse gaussian CDF
X = rank_matrix_to_inverse(X)
# Fit generator
generator.fit(X)
# Generating artificial data \n Sampling from generator
X_artif = generator.sample(n)
# Marginal retrofitting
result = autopandas.AutoData(marginal_retrofit(X_artif, X_real))
# Restore data frame index
result.indexes = indexes
result.columns = columns
return result
def sample(self, n=1, **kwargs):
""" Sample from trained GMM.
:param n: Number of examples to sample.
"""
if self.indexes is None:
raise Exception('You firstly need to train the GMM before sampling. Please use fit method.')
else:
gen_data = self.model.sample(n, **kwargs)[0] # sklearn's gmm return a tuple
return autopandas.AutoData(gen_data, columns=self.columns, indexes=self.indexes)
def sample(self, n=100, loc=0, scale=1):
""" :param scale: Standard deviation of gaussian distribution prior.
"""
randoms = np.array([np.random.normal(loc, scale, self.latent_dim) for _ in range(n)])
decoded = self.decode(randoms)
decoded = autopandas.AutoData(decoded)
if self.columns is not None:
decoded.columns = self.columns
if self.indexes is not None:
decoded.indexes = self.indexes
return decoded
def sample(self, n=100, loc=0, scale=1):
""" :param scale: Standard deviation of gaussian distribution prior.
"""
randoms = np.array(
[np.random.normal(loc, scale, self.latent_dim) for _ in range(n)])
decoded = self.decoder.predict(randoms)
try:
decoded = autopandas.AutoData(decoded)
if self.columns is not None:
decoded.columns = self.columns
if self.indexes is not None:
decoded.indexes = self.indexes
except:
warn('Impossible to cast sampled data to autopandas.AutoData')
return decoded
def decode(new_data, data, limits, min_max):
""" Decode the data from SDV format.
:param data: Data in SDV format
:param data: Original data
:param limits: Limits returned by sdv.encode
:param min_max: Min-max returned by sdv.encode
"""
new_data = autopandas.AutoData(new_data,
columns=data.columns,
indexes=data.indexes)
for c in new_data.columns:
if c in limits:
new_data[c] = undo_categorical(new_data[c], limits[c])
else:
new_data[c] = undo_numeric(new_data[c], *min_max[c])
return new_data
