def bench_ball_tree(N=2000, D=3, k=15, leaf_size=30):
print("Ball Tree")
X = np.random.random((N, D)).astype(DTYPE)
t0 = time()
btskl = skBallTree(X, leaf_size=leaf_size)
t1 = time()
bt = BallTree(X, leaf_size=leaf_size)
t2 = time()
print("Build:")
print(" sklearn : %.2g sec" % (t1 - t0))
print(" new : %.2g sec" % (t2 - t1))
t0 = time()
Dskl, Iskl = btskl.query(X, k)
t1 = time()
dist = [Dskl]
ind = [Iskl]
times = [t1 - t0]
labels = ['sklearn']
counts = [-1]
for dualtree in (False, True):
for breadth_first in (False, True):
bt.reset_n_calls()
t0 = time()
D, I = bt.query(X,
k,
dualtree=dualtree,
breadth_first=breadth_first)
t1 = time()
dist.append(D)
ind.append(I)
times.append(t1 - t0)
counts.append(bt.get_n_calls())
if dualtree:
label = 'dual/'
else:
label = 'single/'
if breadth_first:
label += 'breadthfirst'
else:
label += 'depthfirst'
labels.append(label)
print("Query:")
for lab, t, c in zip(labels, times, counts):
print(" %s : %.2g sec (%i calls)" % (lab, t, c))
print
print(
" distances match: %s" % ', '.join([
'%s' % np.allclose(dist[i - 1], dist[i]) for i in range(len(dist))
]))
print(
" indices match: %s" % ', '.join(
['%s' % np.allclose(ind[i - 1], ind[i]) for i in range(len(ind))]))
def test_tree(N=10000, D=20, K=1, LS=40):
from time import time
from sklearn.neighbors import BallTree as skBallTree
rseed = np.random.randint(10000)
print("-------------------------------------------------------")
print("{0} neighbors of {1} points in {2} dimensions".format(K, N, D))
print("random seed = {0}".format(rseed))
np.random.seed(rseed)
X = np.random.random((N, D))
t0 = time()
bt1 = skBallTree(X, leaf_size=LS)
t1 = time()
dist1, ind1 = bt1.query(X, K)
t2 = time()
bt2 = BallTree(X, leaf_size=LS)
t3 = time()
dist2, ind2 = bt2.query(X, K)
t4 = time()
print("results match: {0} {1}".format(np.allclose(dist1, dist2),
np.allclose(ind1, ind2)))
print("")
print("sklearn build: {0:.2g} sec".format(t1 - t0))
print("python build : {0:.2g} sec".format(t3 - t2))
print("")
print("sklearn query: {0:.2g} sec".format(t2 - t1))
print("python query : {0:.2g} sec".format(t4 - t3))
def bench_ball_tree(N=2000, D=3, k=15, leaf_size=30):
print("Ball Tree")
X = np.random.random((N, D)).astype(DTYPE)
t0 = time()
btskl = skBallTree(X, leaf_size=leaf_size)
t1 = time()
bt = BallTree(X, leaf_size=leaf_size)
t2 = time()
print("Build:")
print(" sklearn : %.2g sec" % (t1 - t0))
print(" new : %.2g sec" % (t2 - t1))
t0 = time()
Dskl, Iskl = btskl.query(X, k)
t1 = time()
dist = [Dskl]
ind = [Iskl]
times = [t1 - t0]
labels = ['sklearn']
counts = [-1]
for dualtree in (False, True):
for breadth_first in (False, True):
bt.reset_n_calls()
t0 = time()
D, I = bt.query(X, k, dualtree=dualtree,
breadth_first=breadth_first)
t1 = time()
dist.append(D)
ind.append(I)
times.append(t1 - t0)
counts.append(bt.get_n_calls())
if dualtree:
label = 'dual/'
else:
label = 'single/'
if breadth_first:
label += 'breadthfirst'
else:
label += 'depthfirst'
labels.append(label)
print("Query:")
for lab, t, c in zip(labels, times, counts):
print(" %s : %.2g sec (%i calls)" % (lab, t, c))
print
print(" distances match: %s"
% ', '.join(['%s' % np.allclose(dist[i - 1], dist[i])
for i in range(len(dist))]))
print(" indices match: %s"
% ', '.join(['%s' % np.allclose(ind[i - 1], ind[i])
for i in range(len(ind))]))
def test_tree(K=2, LS=3):
df = gpd.read_file('../datasets/POINT/UK.geojson')
X = np.stack(df['geometry']).astype(np.float32)
print("-------------------------------------------------------")
print("{0} neighbors of {1} points".format(K, len(X)))
# pre-run to jit compile the code
BallTree(X, leaf_size=LS).query(X, K)
t0 = time()
bt1 = skBallTree(X, leaf_size=LS, metric=vincenty)
t1 = time()
dist1, ind1 = bt1.query(X, K)
t2 = time()
bt2 = BallTree(X, leaf_size=LS)
t3 = time()
# geometry = gpd.GeoSeries(map(Point, bt2.node_centroids))
# df = gpd.GeoDataFrame(geometry=geometry, crs={'init': 'epsg:4326'})
# df['node_idx_start'] = bt2.node_idx[:, 0]
# df['node_idx_end'] = bt2.node_idx[:, 1]
# df['idx'] = np.arange(bt2.n_nodes)
# df['radius'] = bt2.node_radius
# df['geometry'] = df.apply(point_to_circle, axis=1)
# df.to_file('nodes4_.geojson', driver='GeoJSON')
dist2, ind2 = bt2.query(X, K)
t4 = time()
t5 = time()
brute_dist = brute_min(X)
t6 = time()
print('Brute dist = sklearn:', np.allclose(dist1[:, 1], brute_dist))
print('Brute dist = my dist:', np.allclose(dist2[:, 1], brute_dist))
print()
print('My dist = sklearn:', np.allclose(dist1, dist2))
print('My index = sklearn', np.allclose(ind1, ind2, rtol=0))
print()
print("sklearn build: {0:.3g} sec".format(t1 - t0))
print("numba build : {0:.3g} sec".format(t3 - t2))
print()
print("sklearn query: {0:.3g} sec".format(t2 - t1))
print("numba query : {0:.3g} sec".format(t4 - t3))
print("brute query : {0:.3g} sec".format(t6 - t5))
def test_tree(N=1000, D=3, K=5, LS=40):
from time import time
from sklearn.neighbors import BallTree as skBallTree
print("-------------------------------------------------------")
print("Numba version: " + numba.__version__)
rseed = np.random.randint(10000)
print("-------------------------------------------------------")
print("{0} neighbors of {1} points in {2} dimensions".format(K, N, D))
print("random seed = {0}".format(rseed))
np.random.seed(rseed)
X = np.random.random((N, D))
# pre-run to jit compile the code
BallTree(X, leaf_size=LS).query(X, K)
t0 = time()
bt1 = skBallTree(X, leaf_size=LS)
t1 = time()
dist1, ind1 = bt1.query(X, K)
t2 = time()
bt2 = BallTree(X, leaf_size=LS)
t3 = time()
dist2, ind2 = bt2.query(X, K)
t4 = time()
print("results match: {0} {1}".format(np.allclose(dist1, dist2),
np.allclose(ind1, ind2)))
print("")
print("sklearn build: {0:.3g} sec".format(t1 - t0))
print("numba build : {0:.3g} sec".format(t3 - t2))
print("")
print("sklearn query: {0:.3g} sec".format(t2 - t1))
print("numba query : {0:.3g} sec".format(t4 - t3))
from time import time
import numpy as np
from ball_tree import BallTree as pyBallTree
from sklearn.neighbors import BallTree as skBallTree
X = np.random.random((10000, 3))
t0 = time()
pyBT = pyBallTree(X, 30)
t1 = time()
print "py construction: %.2g sec" % (t1 - t0)
t0 = time()
skBT = skBallTree(X, 30)
t1 = time()
print "sk construction: %.2g sec" % (t1 - t0)
for k in [1, 2, 4, 8]:
print "query %i in [%i, %i]:" % (k, X.shape[0], X.shape[1])
t0 = time()
pyBT.query(X, k, dualtree=False)
t1 = time()
print " py: %.2g sec" % (t1 - t0)
t0 = time()
skBT.query(X, k)
t1 = time()
print " sk: %.2g sec" % (t1 - t0)
for r in 0.1, 0.3, 0.5:
from time import time
import numpy as np
from ball_tree import BallTree as pyBallTree
from sklearn.neighbors import BallTree as skBallTree
X = np.random.random((10000, 3))
t0 = time()
pyBT = pyBallTree(X, 30)
t1 = time()
print "py construction: %.2g sec" % (t1 - t0)
t0 = time()
skBT = skBallTree(X, 30)
t1 = time()
print "sk construction: %.2g sec" % (t1 - t0)
for k in [1, 2, 4, 8]:
print "query %i in [%i, %i]:" % (k, X.shape[0], X.shape[1])
t0 = time()
pyBT.query(X, k, dualtree=False)
t1 = time()
print " py: %.2g sec" % (t1 - t0)
t0 = time()
skBT.query(X, k)
t1 = time()
print " sk: %.2g sec" % (t1 - t0)
for r in 0.1, 0.3, 0.5:
