startTime = process_time()
neighbours = vptree.knnQueryBatch(query, k=100, num_threads=2)
end_time = process_time()
searchTime = end_time - startTime
rez = []
dist = []
for i in neighbours:
rez.append(list(i[0]))
dist.append(list(i[1]))
rez = hp.fillIfNotAllAreFound(rez)
result = np.asanyarray(rez)
vptreeRecall = hp.returnRecall(result, groundTruth)
avgDist = np.mean(list(chain.from_iterable(dist)))
reacll.append(vptreeRecall)
algorithm.append('vp-Tree-10k-mL' + str(maxLeave))
#algorithm.append('vp-Tree-maxLeaves'+str(maxLeaves))
construciotnTimes.append(constructionTime)
searchTimes.append(searchTime)
avgdistances.append(avgDist)
del rez
del dist
del result
gc.collect()
#vptree.saveIndex('vptreeIndex.ann')
del vptree
startTime = process_time()
for q in query:
res,d = t.get_nns_by_vector(q, 100, search_k = searchK, include_distances=True)
rez.append(res)
dist.append(d)
#result.append(t.get_nns_by_vector(q, 100, include_distances=True))
end_time = process_time()
searchTime = end_time - startTime
endClock = time.clock()
searchClock= endClock - startClock
result = hp.fillIfNotAllAreFound(rez)
result = np.asanyarray(result)
annoyRecall = hp.returnRecall(result, groundTruth)
avgDist = np.mean(list(chain.from_iterable(dist)))
reacll.append(annoyRecall)
algorithm.append('Annoy-trees-'+str(numTrees))
construciotnTimes.append(constructionTime)
searchTimes.append(searchTime)
avgdistances.append(avgDist)
searchClocks.append(searchClock)
constructionClocks.append(constructionClock)
clockAlg.append('Annoy-trees-'+str(numTrees))
t.save('annoyIndex90.ann')
del t
del rez
del dist
del result
neighbours = hnsw.knnQueryBatch(query, k=100, num_threads=2)
end_time = process_time()
searchTime = end_time - startTime
endClock = time.clock()
searchClock = endClock - startClock
rez = []
dist = []
for i in neighbours:
rez.append(list(i[0]))
dist.append(list(i[1]))
result = hp.fillIfNotAllAreFound(rez)
result = np.array(rez)
hnswRecall = hp.returnRecall(result, groundTruth)
avgDist = np.mean(np.sqrt(list(chain.from_iterable(dist))))
reacll.append(hnswRecall)
algorithm.append('HNSW-M-' + str(MMAX))
construciotnTimes.append(constructionTime)
searchTimes.append(searchTime)
avgdistances.append(avgDist)
constructionClocks.append(constructionClock)
searchClocks.append(searchClock)
clockAlg.append('HNSW-M-' + str(MMAX))
hnsw.saveIndex('hnswIndex40.ann')
del hnsw
del rez
del dist
del result
#flannparams = flann.build_index(train, algorithm ='autotuned', target_precision=tp, build_weight=0 ,memory_weight=0, sample_fraction=0.05)
end_time = process_time()
constructionTime = end_time - startTime
endClock = time.clock()
constructionClock = endClock - startClock
startClock = time.clock()
startTime = process_time()
result, dist = flann.nn_index(query, 100)
end_time = process_time()
searchTime = end_time - startTime
endClock = time.clock()
searchClock = endClock - startClock
result = hp.fillIfNotAllAreFound(result)
result = np.asanyarray(result)
rkdDflannRecall = hp.returnRecall(result, groundTruth)
avgDist = np.mean(np.sqrt(dist))
para.append(flannparams)
reacll.append(rkdDflannRecall)
algorithm.append(flannparams['algorithm'] + '-flann-build005')
construciotnTimes.append(constructionTime)
searchTimes.append(searchTime)
avgdistances.append(avgDist)
searchClocks.append(searchClock)
constructionClocks.append(constructionClock)
clockAlg.append(flannparams['algorithm'] + '-flann-build005')
avgdistances = []
#bruteForce
from sklearn.neighbors import NearestNeighbors
startTime = process_time()
nbrs = NearestNeighbors(n_neighbors=100, algorithm='brute').fit(train)
end_time = process_time()
constructionTime = end_time - startTime
startTime = process_time()
dist, result = nbrs.kneighbors(query, return_distance=True)
end_time = process_time()
searchTime = end_time - startTime
bruteRecall = hp.returnRecall(result, groundTruth)
avgDist = np.mean(np.mean(dist, axis=1))
reacll.append(bruteRecall)
algorithm.append('brute force')
construciotnTimes.append(constructionTime)
searchTimes.append(searchTime)
avgdistances.append(avgDist)
#KDTree test scikit learn
from sklearn.neighbors import KDTree
startTime = process_time()
kdt = KDTree(train, metric='euclidean', leaf_size=2)
end_time = process_time()
constructionTime = end_time - startTime