def produce(self, *, inputs: Inputs, timeout: float = None, iterations: int = None) -> base.CallResult[Outputs]: # inputs: m x n numpy array
if self._fitted:
output = inputs.iloc[:, self._index]
output.metadata = utils.select_columns_metadata(inputs.metadata, columns=self._index)
return CallResult(output)
else:
raise ValueError('Model should be fitted first.')
def produce(self,
*,
inputs: Inputs,
timeout: float = None,
iterations: int = None) -> base.CallResult[Outputs]:
columns_to_use = self._get_columns(inputs.metadata, self.hyperparams)
_logger.debug(f'converting columns: {columns_to_use}')
_logger.debug(
f'converting columns: {inputs.iloc[:, columns_to_use].columns}')
output = inputs.copy()
for col in columns_to_use:
output.iloc[:, col] = pd.to_numeric(output.iloc[:, col])
column_metadata = output.metadata.query(
(metadata_base.ALL_ELEMENTS, col))
semantic_type = column_metadata.get('semantic_types', None)
if 'http://schema.org/Integer' in semantic_type:
output.metadata = output.metadata.update(
(metadata_base.ALL_ELEMENTS, col),
{'structural_type': int})
elif 'http://schema.org/Float' in semantic_type:
output.metadata = output.metadata.update(
(metadata_base.ALL_ELEMENTS, col),
{'structural_type': float})
# What to do with missing values?
# has_missing_value = pd.isnull(output.iloc[:, col]).sum() > 0
if self.hyperparams['drop_non_numeric_columns']:
_logger.debug(
f'dropping columns: {list(np.where(output.dtypes == object)[0])}'
)
_logger.debug(
f'dropping columns: {output.iloc[:, list(np.where(output.dtypes == object)[0])].columns}'
)
# np.where returns int64 instead of int, D3M metadata checks for int
numeric_colum_indices = [
int(x) for x in np.where(output.dtypes != object)[0]
]
output = output.iloc[:, numeric_colum_indices]
output.metadata = utils.select_columns_metadata(
output.metadata, numeric_colum_indices)
return base.CallResult(output)
def _get_new_df(inputs: container.DataFrame, use_cols: list):
metadata = common_utils.select_columns_metadata(
inputs_metadata=inputs.metadata, columns=use_cols)
new_df = inputs.iloc[:, use_cols]
new_df.metadata = metadata
return new_df
