def score(self, trajectory, ground_truth, verbose=False, **kwargs):
segment_id = self.core(trajectory, **kwargs)
h = harmonic_mean(segments=segment_id,
tsid=ground_truth[0],
label=ground_truth[1])
if verbose:
print(
"Scoring " + ','.join('{0}={1!r}'.format(k, kwargs[k])
for k in kwargs), str(h))
return h
def get_best_eps(self,
error_signal,
train_data,
ground_truth,
verbose=False):
ep_a = self.get_eps_candidates(error_signal)
epsa = {}
epsp = {}
for _, ix_ in ep_a:
segmet_id = self.sws_segmentation(dataframe=train_data,
error_array=error_signal,
ep=_,
verbose=verbose)
h = harmonic_mean(segments=segmet_id,
tsid=ground_truth[0],
label=ground_truth[1])
epsa[_] = h
epsp[_] = ix_
epsa_a = None
epsa_k = None
epsa_p = None
for k in epsa.keys():
if epsa_a is None:
epsa_a = epsa[k]
epsa_k = k
epsa_p = epsp[k]
else:
if epsa_a <= epsa[k]:
epsa_a = epsa[k]
epsa_k = k
epsa_p = epsp[k]
if verbose:
print('best eps:', epsa_a, ' for candidate : ', epsa_k, "p:",
epsa_p)
return epsa_a, epsa_k, epsa_p
except:
print("error reading file")
continue
# try:
adbscan.cluster_trajectory_point(tp)
# print("current cluster:",adbscan.current_cluster)
# except Exception as e:
# print("error clustering:",e)
from Trajlib2.SegmentationEvaluation import purity, coverage, harmonic_mean
adbscan.all_eps()
ground_truth = adbscan.get_segment_id()
labels = adbscan.get_ground_truth()
trans_mod = adbscan.get_ground_mode()
segment_id = adbscan.get_segment_id()
print(labels)
print(trans_mod)
print(segment_id)
print(purity(ground_truth_label=trans_mod, generated_segment_id=segment_id))
print(coverage(ground_truth_segment_id=labels,
generated_segment_id=segment_id))
print(harmonic_mean(segments=segment_id, tsid=labels, label=trans_mod))
from Trajlib2.core.traj_reconstruct import plot_df
import pandas as pd
from Trajlib2.databases import load_datasets
ds = load_datasets.load_data_fishing_data(
"/Users/owner/Trajlib2/Trajlib2/databases/fishing")
print(ds[1].shape, len(labels))
plot_df(ds[1], labels[1:], segment_id[1:])
def score(self, X, ground_truth, **kwargs):
segment_id = self._segment_by_time(X, **kwargs)
h = harmonic_mean(ground_truth, segment_id)
#print("Scoring "+','.join('{0}={1!r}'.format(k, kwargs[k]) for k in kwargs),str(h))
return h