def _compute_sparseness(self):
# Compute distance among the OOB and take the avg of their maximum distance
max_distances_starting_from = {}
for (oob1, oob2) in combinations(self.oobs, 2):
# Compute distance between cells
distance = iterative_levenshtein(oob1['interesting segment'],
oob2['interesting segment'])
self.logger.debug("Distance of OOB %s from OOB %s is %.3f",
oob1["test id"], oob2["test id"], distance)
# Update the max values
if oob1['test id'] in max_distances_starting_from.keys():
max_distances_starting_from[oob1['test id']] = max(
max_distances_starting_from[oob1['test id']], distance)
else:
max_distances_starting_from[oob1['test id']] = distance
mean_distance = np.mean([
list(max_distances_starting_from.values())
]) if len(max_distances_starting_from) > 0 else np.NaN
std_dev = np.std([
list(max_distances_starting_from.values())
]) if len(max_distances_starting_from) > 0 else np.NaN
self.logger.debug("Sparseness: Mean: %.3f, StdDev: %3f", mean_distance,
std_dev)
return mean_distance, std_dev
def get_min_distance_from_set(ind, solution):
distances = list()
ind_spine = get_spine(ind)
for road in solution:
road_spine = get_spine(road)
distances.append(iterative_levenshtein(ind_spine, road_spine))
distances.sort()
return distances[0]
def distance(self, other: 'BeamNGMember'):
#TODO
#return frechet_dist(self.sample_nodes, other.sample_nodes)
return iterative_levenshtein(self.sample_nodes, other.sample_nodes)
def distance(self, other: 'BeamNGMember'):
return iterative_levenshtein(self.sample_nodes, other.sample_nodes)