def main():
print("Testing KD Tree...")
test_times = 100
run_time_1 = run_time_2 = 0
for _ in range(test_times):
# Generate dataset randomly
low = 0
high = 100
n_rows = 10000
n_cols = 3
X = gen_data(low, high, n_rows, n_cols)
y = gen_data(low, high, n_rows)
Xi = gen_data(low, high, n_cols)
# Build KD Tree
tree = KDTree()
tree.build_tree(X, y)
# KD Tree Search
start = time()
nd1 = tree.nearest_neighbour_search(Xi)
run_time_1 += time() - start
ret1 = get_euclidean_distance(Xi, nd1.split[0])
# Exhausted search
start = time()
nd2 = exhausted_search(tree, Xi)
run_time_2 += time() - start
ret2 = get_euclidean_distance(Xi, nd2.split[0])
# Compare result
assert ret1 == ret2, "target:%s\nrestult1:%s\nrestult2:%s\ntree:\n%s" % (
str(Xi), str(nd1), str(nd2), str(tree))
print("%d tests passed!" % test_times)
print("KD Tree Search %.2f s" % run_time_1)
print("Exhausted search %.2f s" % run_time_2)
def exhausted_search(tree, Xi):
"""Linear search the nearest neighbour.
Arguments:
tree {KD Tree}
Xi {list} -- [description]
Returns:
Node -- [description]
"""
dist_best = float('inf')
nd_best = None
que = [tree.root]
while que:
nd = que.pop(0)
dist = get_euclidean_distance(Xi, nd.split[0])
if dist < dist_best:
dist_best = dist
nd_best = nd
if nd.left is not None:
que.append(nd.left)
if nd.right is not None:
que.append(nd.right)
return nd_best
def main():
print("Testing KD Tree...")
test_times = 100
run_time_1 = run_time_2 = 0
for _ in range(test_times):
# Generate dataset randomly
low = 0
high = 100
n_rows = 1000
n_cols = 2
X = gen_data(low, high, n_rows, n_cols)
y = gen_data(low, high, n_rows)
Xi = gen_data(low, high, n_cols)
# Build KNN
k = 2
model = KNeighborsBase()
model.fit(X, y, k_neighbors=k)
# KD Tree Search
start = time()
heap = model._knn_search(Xi)
run_time_1 += time() - start
ret1 = [get_euclidean_distance(Xi, nd.split[0]) for nd in heap.items]
ret1.sort()
# Exhausted search
start = time()
ret2 = exhausted_search(X, Xi, k)
run_time_2 += time() - start
ret2 = [get_euclidean_distance(Xi, row) for row in ret2]
ret2.sort()
# Compare result
assert ret1 == ret2, "target:%s\nrestult1:%s\nrestult2:%s\ntree:\n%s" \
% (Xi, ret1, ret2, model.tree)
print("%d tests passed!" % test_times)
print("KNN Search %.2f s" % run_time_1)
print("Exhausted search %.2f s" % run_time_2)
def exhausted_search(X, Xi):
"""Linear search the nearest neighbour.
Arguments:
X {list} -- 2d list with int or float.
Xi {list} -- 1d list with int or float.
Returns:
list -- 1d list with int or float.r.
"""
dist_best = float('inf')
row_best = None
for row in X:
dist = get_euclidean_distance(Xi, row)
if dist < dist_best:
dist_best = dist
row_best = row
return row_best
def exhausted_search(X, Xi, k):
"""Linear search the nearest neighbour.
Arguments:
X {list} -- 2d list with int or float.
Xi {list} -- 1d list with int or float.
k {int} -- number of neighbours.
Returns:
list -- The lists of the K nearest neighbour.
"""
idxs = []
for _ in range(k):
best_dist = float('inf')
idxs.append(None)
for i, row in enumerate(X):
dist = get_euclidean_distance(Xi, row)
if dist < best_dist and i not in idxs:
best_dist = dist
idxs[-1] = i
return [X[i] for i in idxs]