def describe_pipeline(self, pipeline_id, return_dict=False):
"""Describe a pipeline
Arguments:
pipeline_id (int): pipeline to describe
return_dict (bool): If True, return dictionary of information
about pipeline. Defaults to False.
Returns:
Description of specified pipeline. Includes information such as
type of pipeline components, problem, training time, cross validation, etc.
"""
if pipeline_id not in self._results['pipeline_results']:
raise PipelineNotFoundError("Pipeline not found")
pipeline = self.get_pipeline(pipeline_id)
pipeline_results = self._results['pipeline_results'][pipeline_id]
pipeline.describe()
log_subtitle(logger, "Training")
logger.info("Training for {} problems.".format(pipeline.problem_type))
if self.optimize_thresholds and self.objective.is_defined_for_problem_type(
ProblemTypes.BINARY) and self.objective.can_optimize_threshold:
logger.info(
"Objective to optimize binary classification pipeline thresholds for: {}"
.format(self.objective))
logger.info("Total training time (including CV): %.1f seconds" %
pipeline_results["training_time"])
log_subtitle(logger, "Cross Validation", underline="-")
all_objective_scores = [
fold["all_objective_scores"]
for fold in pipeline_results["cv_data"]
]
all_objective_scores = pd.DataFrame(all_objective_scores)
for c in all_objective_scores:
if c in ["# Training", "# Validation"]:
all_objective_scores[c] = all_objective_scores[c].astype(
"object")
continue
mean = all_objective_scores[c].mean(axis=0)
std = all_objective_scores[c].std(axis=0)
all_objective_scores.loc["mean", c] = mean
all_objective_scores.loc["std", c] = std
all_objective_scores.loc[
"coef of var", c] = std / mean if abs(mean) > 0 else np.inf
all_objective_scores = all_objective_scores.fillna("-")
with pd.option_context('display.float_format', '{:.3f}'.format,
'expand_frame_repr', False):
logger.info(all_objective_scores)
if return_dict:
return pipeline_results
def get_pipeline(self, pipeline_id):
"""Given the ID of a pipeline training result, returns an untrained instance of the specified pipeline
initialized with the parameters used to train that pipeline during automl search.
Arguments:
pipeline_id (int): pipeline to retrieve
Returns:
PipelineBase: untrained pipeline instance associated with the provided ID
"""
pipeline_results = self.results['pipeline_results'].get(pipeline_id)
if pipeline_results is None:
raise PipelineNotFoundError("Pipeline not found in automl results")
pipeline_class = pipeline_results.get('pipeline_class')
parameters = pipeline_results.get('parameters')
if pipeline_class is None or parameters is None:
raise PipelineNotFoundError("Pipeline class or parameters not found in automl results")
return pipeline_class(parameters, random_seed=self.random_seed)
def best_pipeline(self):
"""Returns a trained instance of the best pipeline and parameters found during automl search. If `train_best_pipeline` is set to False, returns an untrained pipeline instance.
Returns:
PipelineBase: A trained instance of the best pipeline and parameters found during automl search. If `train_best_pipeline` is set to False, returns an untrained pipeline instance.
"""
if not self._best_pipeline:
raise PipelineNotFoundError("automl search must be run before selecting `best_pipeline`.")
return self._best_pipeline
def add_result(self, score_to_minimize, pipeline, trained_pipeline_results):
"""Register results from evaluating a pipeline
Arguments:
score_to_minimize (float): The score obtained by this pipeline on the primary objective, converted so that lower values indicate better pipelines.
pipeline (PipelineBase): The trained pipeline object which was used to compute the score.
trained_pipeline_results (dict): Results from training a pipeline.
"""
if pipeline.name not in self._tuners:
raise PipelineNotFoundError(f"No such pipeline allowed in this AutoML search: {pipeline.name}")
self._tuners[pipeline.name].add(pipeline.parameters, score_to_minimize)