Precision annoy.AnnoyIndex with examples#
An example showing the AnnoyIndex class.
from __future__ import print_function
import random; random.seed(0)
import time
# from annoy import AnnoyIndex
# from scikitplot.annoy import AnnoyIndex
from scikitplot.annoy import Index as AnnoyIndex
try:
from tqdm.auto import tqdm, trange
except ImportError:
# Fallback: dummy versions that ignore all args/kwargs
tqdm = lambda iterable, *args, **kwargs: iterable
trange = lambda n, *args, **kwargs: range(n)
n, f = 1_000_000, 100 # 100~2.5GB
n, f = 100_000, 100 # 100~0.25GB 256~0.6GB
idx = AnnoyIndex(
f=f,
metric='angular',
)
idx.set_seed(0)
for i in trange(n):
if(i % (n//10) == 0): print(f"{i} / {n} = {1.0 * i / n}")
# v = []
# for z in range(f):
# v.append(random.gauss(0, 1))
v = [random.gauss(0, 1) for _ in range(f)]
idx.add_item(i, v)
idx.build(2 * f)
idx.save('test.annoy')
idx.info()
/home/circleci/repo/galleries/examples/annoy/plot_precision_script.py:37: UserWarning: seed=0 resets to Annoy's default seed
idx.set_seed(0)
0%| | 0/100000 [00:00<?, ?it/s]0 / 100000 = 0.0
2%|▏ | 1727/100000 [00:00<00:05, 17264.30it/s]
3%|▎ | 3470/100000 [00:00<00:05, 17360.77it/s]
5%|▌ | 5207/100000 [00:00<00:05, 17341.47it/s]
7%|▋ | 6942/100000 [00:00<00:05, 17311.26it/s]
9%|▊ | 8674/100000 [00:00<00:05, 17251.41it/s]10000 / 100000 = 0.1
10%|█ | 10400/100000 [00:00<00:05, 17189.96it/s]
12%|█▏ | 12126/100000 [00:00<00:05, 17194.63it/s]
14%|█▍ | 13871/100000 [00:00<00:04, 17275.00it/s]
16%|█▌ | 15603/100000 [00:00<00:04, 17287.00it/s]
17%|█▋ | 17332/100000 [00:01<00:04, 17187.16it/s]
19%|█▉ | 19067/100000 [00:01<00:04, 17234.03it/s]20000 / 100000 = 0.2
21%|██ | 20792/100000 [00:01<00:04, 17238.55it/s]
23%|██▎ | 22521/100000 [00:01<00:04, 17253.30it/s]
24%|██▍ | 24268/100000 [00:01<00:04, 17315.92it/s]
26%|██▌ | 26000/100000 [00:01<00:04, 17311.91it/s]
28%|██▊ | 27732/100000 [00:01<00:04, 16779.54it/s]
29%|██▉ | 29464/100000 [00:01<00:04, 16936.16it/s]30000 / 100000 = 0.3
31%|███ | 31198/100000 [00:01<00:04, 17053.60it/s]
33%|███▎ | 32941/100000 [00:01<00:03, 17164.56it/s]
35%|███▍ | 34660/100000 [00:02<00:03, 17129.69it/s]
36%|███▋ | 36391/100000 [00:02<00:03, 17180.97it/s]
38%|███▊ | 38121/100000 [00:02<00:03, 17215.03it/s]
40%|███▉ | 39856/100000 [00:02<00:03, 17253.74it/s]40000 / 100000 = 0.4
42%|████▏ | 41588/100000 [00:02<00:03, 17273.09it/s]
43%|████▎ | 43323/100000 [00:02<00:03, 17296.02it/s]
45%|████▌ | 45053/100000 [00:02<00:03, 17109.73it/s]
47%|████▋ | 46789/100000 [00:02<00:03, 17181.61it/s]
49%|████▊ | 48523/100000 [00:02<00:02, 17227.79it/s]50000 / 100000 = 0.5
50%|█████ | 50256/100000 [00:02<00:02, 17255.88it/s]
52%|█████▏ | 51992/100000 [00:03<00:02, 17285.46it/s]
54%|█████▎ | 53728/100000 [00:03<00:02, 17306.04it/s]
55%|█████▌ | 55462/100000 [00:03<00:02, 17314.84it/s]
57%|█████▋ | 57194/100000 [00:03<00:02, 17316.17it/s]
59%|█████▉ | 58926/100000 [00:03<00:02, 17076.60it/s]60000 / 100000 = 0.6
61%|██████ | 60660/100000 [00:03<00:02, 17152.27it/s]
62%|██████▏ | 62392/100000 [00:03<00:02, 17201.65it/s]
64%|██████▍ | 64143/100000 [00:03<00:02, 17291.50it/s]
66%|██████▌ | 65881/100000 [00:03<00:01, 17315.50it/s]
68%|██████▊ | 67615/100000 [00:03<00:01, 17321.77it/s]
69%|██████▉ | 69356/100000 [00:04<00:01, 17345.22it/s]70000 / 100000 = 0.7
71%|███████ | 71094/100000 [00:04<00:01, 17355.27it/s]
73%|███████▎ | 72830/100000 [00:04<00:01, 17345.55it/s]
75%|███████▍ | 74565/100000 [00:04<00:01, 17327.36it/s]
76%|███████▋ | 76298/100000 [00:04<00:01, 17014.76it/s]
78%|███████▊ | 78030/100000 [00:04<00:01, 17102.37it/s]
80%|███████▉ | 79761/100000 [00:04<00:01, 17163.14it/s]80000 / 100000 = 0.8
81%|████████▏ | 81494/100000 [00:04<00:01, 17211.63it/s]
83%|████████▎ | 83230/100000 [00:04<00:00, 17253.74it/s]
85%|████████▍ | 84965/100000 [00:04<00:00, 17280.30it/s]
87%|████████▋ | 86696/100000 [00:05<00:00, 17289.03it/s]
88%|████████▊ | 88434/100000 [00:05<00:00, 17314.25it/s]90000 / 100000 = 0.9
90%|█████████ | 90167/100000 [00:05<00:00, 17315.73it/s]
92%|█████████▏| 91903/100000 [00:05<00:00, 17328.53it/s]
94%|█████████▎| 93642/100000 [00:05<00:00, 17346.70it/s]
95%|█████████▌| 95377/100000 [00:05<00:00, 17343.31it/s]
97%|█████████▋| 97112/100000 [00:05<00:00, 17336.43it/s]
99%|█████████▉| 98846/100000 [00:05<00:00, 17328.97it/s]
100%|██████████| 100000/100000 [00:05<00:00, 17219.32it/s]
{'f': 100, 'metric': 'angular', 'n_neighbors': 5, 'on_disk_path': 'test.annoy', 'prefault': False, 'seed': None, 'verbose': None, 'schema_version': 0, 'n_items': 100000, 'n_trees': 200, 'memory_usage_byte': 499984128, 'memory_usage_mib': 476.822021484375}
def plot(idx, y=None, **kwargs):
import numpy as np
import matplotlib.pyplot as plt
import scikitplot.cexternals._annoy._plotting as utils
single = np.zeros(idx.get_n_items(), dtype=int)
if y is None:
double = np.random.uniform(0, 1, idx.get_n_items()).round()
# single vs double
fig, ax = plt.subplots(ncols=2, figsize=(12, 5))
alpha = kwargs.pop("alpha", 0.8)
y2 = utils.plot_annoy_index(
idx,
dims = list(range(idx.f)),
plot_kwargs={"draw_legend": False},
ax=ax[0],
)[0]
utils.plot_annoy_knn_edges(
idx,
y2,
k=1,
line_kwargs={"alpha": alpha},
ax=ax[1],
)
# idx.unbuild()
# idx.build(10)
plot(idx)
def precision(q):
limits = [10, 100, 1_000, 10_000]
k = 10
prec_n = 10
prec_sum = {}
time_sum = {}
for i in trange(prec_n):
j = random.randrange(0, n)
closest = set(q.get_nns_by_item(j, k, n))
for limit in limits:
t0 = time.time()
toplist = q.get_nns_by_item(j, k, limit)
T = time.time() - t0
found = len(closest.intersection(toplist))
hitrate = 1.0 * found / k
prec_sum[limit] = prec_sum.get(limit, 0.0) + hitrate
time_sum[limit] = time_sum.get(limit, 0.0) + T
for limit in limits:
print('limit: %-9d precision: %6.2f%% avg time: %.6fs'
% (limit, 100.0 * prec_sum[limit] / (i + 1), time_sum[limit] / (i + 1)))
q = AnnoyIndex(f, 'angular')
q.set_seed(0)
q.load('test.annoy')
precision(q)
/home/circleci/repo/galleries/examples/annoy/plot_precision_script.py:111: UserWarning: seed=0 resets to Annoy's default seed
q.set_seed(0)
0%| | 0/10 [00:00<?, ?it/s]
60%|██████ | 6/10 [00:00<00:00, 55.21it/s]
100%|██████████| 10/10 [00:00<00:00, 56.63it/s]
limit: 10 precision: 10.00% avg time: 0.000145s
limit: 100 precision: 16.00% avg time: 0.000119s
limit: 1000 precision: 33.00% avg time: 0.000376s
limit: 10000 precision: 86.00% avg time: 0.002363s
Total running time of the script: (2 minutes 6.031 seconds)
Related examples
