def __init__(self, data):
'''
Reads the data, standardizes it and creates an index based on the data.
Args:
data (array): A two-dimensional array, with vectors to be indexed in the rows.
Preferably a numpy array, but other iterables (e.g. list of lists of the same length) could work too.
Raises:
ValueError, TypeError or something like that, if the dataset is invalid.
'''
standardized_data = standardize(data)
self._tree = KDTree(standardized_data) # Index the data
def _setup_tree(self):
''' sets up the kD-trees and retrieves the min time stamp '''
if not self._observations is None:
timed_obs = defaultdict(list)
for obs in self._observations:
timed_obs[int(obs[2])].append(obs)
self._trees = dict()
for key, val in timed_obs.items():
self._trees[key] = KDTree(val)
self._min_time_stamp = min(timed_obs.keys())
else:
raise Exception("observations not set.")
class CorrelationIndex:
'''
Data structure, that indexes a set of vectors for fast retrieval of
elements most correlated with the query point.
>> fci = CorrelationIndex([(1,3,5), (4,3,2)])
>> fci.query((1,2,3))
[0]
>> fci.query((3,2,1))
[1]
>> fci.query((1,2,3), 0)
[]
>> fci.query((1,2,3), 2)
[0, 1]
>> fci.query((3,2,1), 3)
[1, 0]
>> fci = CorrelationIndex([])
>> fci.query((1,2,3))
[]
>> fci = CorrelationIndex("random string")
Traceback (most recent call last):
...
ValueError: object is not iterable
>> fci = CorrelationIndex([(1,3,5), (4,3)])
Traceback (most recent call last):
...
ValueError: iterable should contain tuples (or vectors or lists) of numbers of equal length
'''
def __init__(self, data):
'''
Reads the data, standardizes it and creates an index based on the data.
Args:
data (array): A two-dimensional array, with vectors to be indexed in the rows.
Preferably a numpy array, but other iterables (e.g. list of lists of the same length) could work too.
Raises:
ValueError, TypeError or something like that, if the dataset is invalid.
'''
standardized_data = standardize(data)
self._tree = KDTree(standardized_data) # Index the data
def query(self, x, matches=1):
#TODO: Implement
return [self._tree.find_approximate_nearest_neigbor(x)]
