-
Notifications
You must be signed in to change notification settings - Fork 3
/
predictions.py
79 lines (62 loc) · 2.68 KB
/
predictions.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
import typing
from argparse import ArgumentParser
from collections import Counter
import pandas as pd
def has_majority(counter_value: list[tuple[typing.Any, int]]) -> bool:
if len(counter_value) == 1:
return True
else:
return counter_value[0][1] > counter_value[1][1]
def prediction_vote(dfs: list[pd.DataFrame], delete_if_no_majority: bool) -> pd.DataFrame:
"""
Combine multiple prediction results by voting.
:param dfs: List of DataFrames with prediction results
:param delete_if_no_majority: Should websites be deleted from the predictions if there is not just one label that
has the highest number of votes.
:return: DataFrame where the predicted_label or decision column is decided by voting
"""
key = "predicted_label" if "predicted_label" in dfs[0] else "decision"
class_votes: list[list[str | bool]] = [df[key].tolist() for df in dfs]
top_choices: list[list[tuple[str | bool, int]]] = [Counter(x).most_common(2) for x in zip(*class_votes)]
if delete_if_no_majority:
has_maj = [has_majority(c) for c in top_choices]
majority_votes = [c[0][0] for c in top_choices if has_majority(c)]
df = dfs[0][pd.Series(has_maj)].reset_index(drop=True)
df[key] = majority_votes
else:
df = dfs[0].copy()
df[key] = [c[0][0] for c in top_choices]
if key == "decision":
df.drop("prediction", axis="columns", inplace=True)
if "entropy" in df:
df.drop("entropy", axis="columns", inplace=True)
return df
def print_distribution(df: pd.DataFrame):
c = Counter(df["predicted_label" if "predicted_label" in df else "decision"])
total = c.total()
for category, count in c.most_common():
print(f"{category};{count};{count * 100 / total:.2f}%")
def main():
parser = ArgumentParser()
sp = parser.add_subparsers(dest="command")
sp.required = True
sp_combine = sp.add_parser("combine")
sp_combine.add_argument("out")
sp_combine.add_argument("p1")
sp_combine.add_argument("p2")
sp_combine.add_argument("p3")
sp_combine.add_argument("--delete-if-no-majority", action="store_true")
sp_distr = sp.add_parser("distribution")
sp_distr.add_argument("predictions")
args = parser.parse_args()
if args.command == "combine":
df1 = pd.read_parquet(args.p1)
df2 = pd.read_parquet(args.p2)
df3 = pd.read_parquet(args.p3)
combined = prediction_vote([df1, df2, df3], args.delete_if_no_majority)
combined.to_parquet(args.out)
elif args.command == "distribution":
df = pd.read_parquet(args.predictions)
print_distribution(df)
if __name__ == "__main__":
main()