def search(dataset_root, problem, args):
pps = PipelineSearcher(
args.input,
args.output,
args.static,
dump=True,
hard_timeout=args.hard,
)
return pps.search(problem, args.timeout, args.budget, args.template)
def search(dataset_root, problem, args):
pps = PipelineSearcher(args.input, args.output, dump=True)
return pps.search(problem, timeout=args.timeout, budget=args.budget)
def process_dataset(dataset_name, dataset, problem, args):
box_print("Processing dataset {}".format(dataset_name), True)
output_path = os.path.join(args.output, dataset_name)
os.makedirs(output_path, exist_ok=True)
LOGGER.info("Searching Pipeline for dataset {}".format(dataset_name))
try:
start_ts = datetime.utcnow()
pps = PipelineSearcher(args.input,
output_path,
args.static,
dump=True,
hard_timeout=args.hard,
ignore_errors=args.ignore_errors,
cv_folds=args.folds,
subprocess_timeout=args.subprocess_timeout,
max_errors=args.max_errors,
store_summary=True)
result = pps.search(dataset, problem, args.timeout, args.budget,
args.templates_csv)
result['elapsed'] = datetime.utcnow() - start_ts
result['dataset'] = dataset_name
except Exception as ex:
result = {
'dataset': dataset_name,
'error': '{}: {}'.format(type(ex).__name__, ex),
}
else:
try:
summary = result.pop('summary')
candidates = _select_candidates(summary)
if candidates.empty:
box_print('No valid pipelines found for dataset {}'.format(
dataset_name))
else:
ranked_path = os.path.join(output_path, 'pipelines_ranked')
test_scores = list()
for _, candidate in candidates.iterrows():
try:
pipeline = candidate.pipeline
pipeline_path = os.path.join(ranked_path, pipeline)
test_score = score_pipeline(dataset, problem,
pipeline_path, args.static,
output_path)
test_scores.append(test_score)
except Exception:
test_scores.append(None)
candidates['test_score'] = test_scores
candidates = candidates.sort_values('test_score',
ascending=False)
best = candidates.iloc[0]
result['test_score'] = best.test_score
result['template'] = best.template
result['cv_score'] = best.score
box_print('Best pipelines for dataset {}:\n{}'.format(
dataset_name, candidates.to_string()))
except Exception as ex:
LOGGER.exception('Error while testing the winner pipeline')
result['error'] = 'TEST Error: {}'.format(ex)
return result
