def init(clustering: str = ClusteringMethods.NO_CLUSTER.value,
configuration: dict = {}):
if clustering == ClusteringMethods.NO_CLUSTER.value:
return NoCluster.objects.get_or_create(
pk=1, clustering_method=clustering)[0]
elif clustering == ClusteringMethods.KMEANS.value:
from src.clustering.methods_default_config import clustering_kmeans # TODO fixme
default_configuration = clustering_kmeans()
return KMeans.objects.get_or_create(
clustering_method=clustering,
n_clusters=configuration.get(
'n_clusters', default_configuration['n_clusters']),
init=configuration.get('init', default_configuration['init']),
n_init=configuration.get('n_init',
default_configuration['n_init']),
max_iter=configuration.get('max_iter',
default_configuration['max_iter']),
tol=configuration.get('tol', default_configuration['tol']),
precompute_distances=configuration.get(
'precompute_distances',
default_configuration['precompute_distances']),
random_state=configuration.get(
'random_state', default_configuration['random_state']),
copy_x=configuration.get('copy_x',
default_configuration['copy_x']),
algorithm=configuration.get(
'algorithm', default_configuration['algorithm']))[0]
else:
raise ValueError(
'configuration {} not recognized'.format(clustering))
def add_default_config(job: Job, prediction_method=""): # TODO is this needed?
"""Map to job method default config"""
if prediction_method == "":
prediction_method = job.predictive_model.prediction_method
method_conf_name = "{}.{}".format(job.predictive_model.predictive_model,
prediction_method)
method_conf = CONF_MAP[method_conf_name]()
job[method_conf_name] = method_conf
job['kmeans'] = clustering_kmeans()
return job