forked from nyu-dl/dl4marco-bert
-
Notifications
You must be signed in to change notification settings - Fork 0
/
metrics.py
93 lines (73 loc) · 2.07 KB
/
metrics.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
import numpy as np
def average_precision(gt, pred):
"""
Computes the average precision.
This function computes the average prescision at k between two lists of
items.
Parameters
----------
gt: set
A set of ground-truth elements (order doesn't matter)
pred: list
A list of predicted elements (order does matter)
Returns
-------
score: double
The average precision over the input lists
"""
if not gt:
return 0.0
score = 0.0
num_hits = 0.0
for i,p in enumerate(pred):
if p in gt and p not in pred[:i]:
num_hits += 1.0
score += num_hits / (i + 1.0)
return score / max(1.0, len(gt))
def NDCG(gt, pred, use_graded_scores=False):
score = 0.0
for rank, item in enumerate(pred):
if item in gt:
if use_graded_scores:
grade = 1.0 / (gt.index(item) + 1)
else:
grade = 1.0
score += grade / np.log2(rank + 2)
norm = 0.0
for rank in range(len(gt)):
if use_graded_scores:
grade = 1.0 / (rank + 1)
else:
grade = 1.0
norm += grade / np.log2(rank + 2)
return score / max(0.3, norm)
def metrics(gt, pred, metrics_map):
'''
Returns a numpy array containing metrics specified by metrics_map.
gt: ground-truth items
pred: predicted items
'''
out = np.zeros((len(metrics_map),), np.float32)
if ('MAP' in metrics_map):
avg_precision = average_precision(gt=gt, pred=pred)
out[metrics_map.index('MAP')] = avg_precision
if ('RPrec' in metrics_map):
intersec = len(gt & set(pred[:len(gt)]))
out[metrics_map.index('RPrec')] = intersec / max(1., float(len(gt)))
if 'MRR' in metrics_map:
score = 0.0
for rank, item in enumerate(pred):
if item in gt:
score = 1.0 / (rank + 1.0)
break
out[metrics_map.index('MRR')] = score
if 'MRR@10' in metrics_map:
score = 0.0
for rank, item in enumerate(pred[:10]):
if item in gt:
score = 1.0 / (rank + 1.0)
break
out[metrics_map.index('MRR@10')] = score
if ('NDCG' in metrics_map):
out[metrics_map.index('NDCG')] = NDCG(gt, pred)
return out