def _transform(self, row):
"""Transforms the sparse vector to a dense vector while putting it in a new column."""
sparse_vector = row[self.input_column]
dense_vector = DenseVector(sparse_vector.toArray())
new_row = new_dataframe_row(row, self.output_column, dense_vector)
return new_row
def _transform(self, row):
"""Transforms every row by adding a "predicted index" column to the dataframe. """
prediction = row[self.input_column]
index = float(self.get_index(prediction))
new_row = new_dataframe_row(row, self.output_column, index)
return new_row
def _transform(self, row):
"""Transforms the vector to a dense matrix while putting it in a new column."""
vector = row[self.input_column]
reshaped = vector.toArray().reshape(self.shape).tolist()
new_row = new_dataframe_row(row, self.output_column, reshaped)
return new_row
def _transform(self, row):
"""Transforms the vector to a dense matrix while putting it in a new column."""
vector = row[self.input_column]
vector = np.asarray(vector)
reshaped = vector.reshape(self.shape).tolist()
new_row = new_dataframe_row(row, self.output_column, reshaped)
return new_row
def _transform(self, row):
"""Transforms every individual row.
Only for internal use.
"""
label = row[self.input_column]
vector = to_one_hot_encoded_dense(label, self.output_dimensionality)
new_row = new_dataframe_row(row, self.output_column, vector.tolist())
return new_row
def _transform(self, row):
"""Take the column, and normalize it with the computed means and std devs."""
mean = self.means[self.current_column]
stddev = self.stddevs[self.current_column]
x = row[self.current_column]
x_normalized = (x - mean) / stddev
output_column = self.current_column + self.column_suffix
new_row = new_dataframe_row(row, output_column, x_normalized)
return new_row
def _transform(self, row):
"""Appends the desired binary label column."""
value = row[self.input_column]
vector = np.zeros(2)
# Check if the name matches.
if value == self.label:
vector[0] = 1.0
else:
vector[1] = 1.0
# Convert to a Spark DenseVector
vector = DenseVector(vector)
return new_dataframe_row(row, self.output_column, vector)
def _transform(self, row):
"""Rescale every instance like this:
x' = \frac{x - min}{max - min}
"""
vector = row[self.input_column].toArray()
vector = self.scale * (vector - self.o_max) + self.n_max
# Convert to a DenseVector.
dense_vector = DenseVector(vector)
# Construct a new row with the normalized vector.
new_row = new_dataframe_row(row, self.output_column, dense_vector)
return new_row
def _predict(self, iterator):
"""Lambda method which will append a prediction column to the provided rows.
# Arguments:
iterator: iterator. Spark Row iterator.
"""
model = deserialize_keras_model(self.model)
for row in iterator:
features = [np.asarray([row[c]]) for c in self.features_column]
prediction = model.predict(features)
dense_prediction = DenseVector(prediction[0])
new_row = new_dataframe_row(row, self.output_column, dense_prediction)
yield new_row
def _predict(self, iterator):
"""Lambda method which will append a prediction column to the provided rows.
# Arguments:
iterator: iterator. Spark Row iterator.
"""
model = deserialize_keras_model(self.model)
for row in iterator:
features = np.asarray([row[self.features_column]])
prediction = model.predict(features)
dense_prediction = DenseVector(prediction[0])
new_row = new_dataframe_row(row, self.output_column,
dense_prediction)
yield new_row
def _transform(self, row):
"""Rescale every instance like this:
x' = \frac{x - min}{max - min}
"""
if self.is_vector:
vector = row[self.input_column].toArray()
vector = self.scale * (vector - self.o_max) + self.n_max
new_value = DenseVector(vector)
else:
value = row[self.input_column]
new_value = self.scale * (value - self.o_max) + self.n_max
# Construct a new row with the normalized vector.
new_row = new_dataframe_row(row, self.output_column, new_value)
return new_row
def _transform(self, row):
"""Transforms every individual row.
Only for internal use.
"""
label = row[self.input_column]
if (isinstance(label, types.ListType)):
vector = np.zeros((len(label), self.output_dimensionality))
for i in range(len(label)):
vector[i] = to_one_hot_encoded_dense(
label[i], self.output_dimensionality)
else:
vector = to_one_hot_encoded_dense(label,
self.output_dimensionality)
new_row = new_dataframe_row(row, self.output_column, vector.tolist())
return new_row
