-
Notifications
You must be signed in to change notification settings - Fork 2
/
09_vote_ensemble.py
142 lines (116 loc) · 4.88 KB
/
09_vote_ensemble.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
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
#=======================================================================================================================
# This file generates an ensemble prediction using outputs from 06, 07, 08 scripts
# predictions will be saved in wdata/submits folder
# threshold = 4 would be valid choice according to out-of-fold predictions
# @ Shaikat Galib / [email protected] / 25/04/2019
#=======================================================================================================================
from collections import Counter
import numpy as np
import pandas as pd
import sys
#=======================================================================================================================
# bash test.sh /data/testing/ solution.csv
test_folder = '/data/testing/'
solution_fn = 'solution.csv'
wdata_dir = '/wdata/'
if len(sys.argv) > 1:
test_folder = sys.argv[1]
solution_fn = sys.argv[2]
#======================================================================================================================
# folder name to save submits
expt_name = 'ann'
sub_dir = wdata_dir + 'submits/' + expt_name + '/'
threshold = 4 # oof best 4
# df1 = pd.read_csv(sub_dir + 'solution_{}_th3_seg0.5_test.csv'.format(expt_name))
# df2 = pd.read_csv(sub_dir + 'solution_{}_th5_seg0.5_test.csv'.format(expt_name))
# df3 = pd.read_csv(sub_dir + 'solution_{}_th7_seg0.5_test.csv'.format(expt_name))
df4 = pd.read_csv(sub_dir + 'solution_{}_th3_seg1.25_test.csv'.format(expt_name))
df5 = pd.read_csv(sub_dir + 'solution_{}_th5_seg1.25_test.csv'.format(expt_name))
df6 = pd.read_csv(sub_dir + 'solution_{}_th7_seg1.25_test.csv'.format(expt_name))
# df7 = pd.read_csv(sub_dir + 'solution_{}_th3_seg0.33_test.csv'.format(expt_name))
# df8 = pd.read_csv(sub_dir + 'solution_{}_th5_seg0.33_test.csv'.format(expt_name))
# df9 = pd.read_csv(sub_dir + 'solution_{}_th7_seg0.33_test.csv'.format(expt_name))
df10 = pd.read_csv(sub_dir + 'solution_{}_th3_seg1500_test.csv'.format(expt_name))
df11 = pd.read_csv(sub_dir + 'solution_{}_th5_seg1500_test.csv'.format(expt_name))
df12 = pd.read_csv(sub_dir + 'solution_{}_th7_seg1500_test.csv'.format(expt_name))
df13 = pd.read_csv(sub_dir + 'solution_{}_th3_seg3000_test.csv'.format(expt_name))
df14 = pd.read_csv(sub_dir + 'solution_{}_th5_seg3000_test.csv'.format(expt_name))
df15 = pd.read_csv(sub_dir + 'solution_{}_th7_seg3000_test.csv'.format(expt_name))
ids = df4.RunID
sources = df4.SourceID
df_combined = pd.DataFrame()
df_combined['RunID'] = ids
# sid1 = df1.SourceID.values
# sid2 = df2.SourceID.values
# sid3 = df3.SourceID.values
sid4 = df4.SourceID.values
sid5 = df5.SourceID.values
sid6 = df6.SourceID.values
# sid7 = df7.SourceID.values
# sid8 = df8.SourceID.values
# sid9 = df9.SourceID.values
sid10 = df10.SourceID.values
sid11 = df11.SourceID.values
sid12 = df12.SourceID.values
sid13 = df13.SourceID.values
sid14 = df14.SourceID.values
sid15 = df15.SourceID.values
# time1 = df1.SourceTime.values
# time2 = df2.SourceTime.values
# time3 = df3.SourceTime.values
time4 = df4.SourceTime.values
time5 = df5.SourceTime.values
time6 = df6.SourceTime.values
# time7 = df7.SourceTime.values
# time8 = df8.SourceTime.values
# time9 = df9.SourceTime.values
time10 = df10.SourceTime.values
time11 = df11.SourceTime.values
time12 = df12.SourceTime.values
time13 = df13.SourceTime.values
time14 = df14.SourceTime.values
time15 = df15.SourceTime.values
np_sid = np.array([#sid1, sid2, sid3,
sid4, sid5, sid6,
#sid7, sid8, sid9,
sid10, sid11, sid12,
sid13, sid14, sid15
]).T
np_time = np.array([#time1, time2, time3,
time4, time5, time6,
#time7, time8, time9,
time10, time11, time12,
time13, time14, time15
, ], dtype=np.float16).T
run_id = []
filtered_label = []
filtered_time = []
for i, rid in enumerate(ids):
all_labels = np_sid[i]
all_timess = np_time[i]
# Count the frequency of non-zero preds
try:
preds_nonzero = [x for x in all_labels if x > 0]
most_common, freq_most_common = Counter(preds_nonzero).most_common(1)[0]
except:
freq_most_common = 0
#most_common, freq_most_common = Counter(all_labels).most_common(1)[0]
if freq_most_common >= threshold:
idx = np.where(all_labels==most_common)
_time = all_timess[idx]
avg_time = sum(_time)/len(_time)
run_id.append(rid)
filtered_label.append(most_common)
filtered_time.append(avg_time)
#print('done')
else:
run_id.append(rid)
filtered_label.append(0)
filtered_time.append(0)
sub = pd.DataFrame()
sub["RunID"] = run_id
sub['SourceID'] = filtered_label
sub["SourceTime"] = filtered_time
print(sub['SourceID'].astype(bool).sum(axis=0))
sub.to_csv(sub_dir + "{}_3tta_th{}_test.csv".format(expt_name, threshold), index=False)
print('done')