-
Notifications
You must be signed in to change notification settings - Fork 4
/
utility.py
102 lines (82 loc) · 2.96 KB
/
utility.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
import numpy as np
import pandas as pd
import sklearn
import json
import re
def get_stage2_train_data(preprocessed=False):
train = pd.read_csv('data/train_stage2.csv')
y = train['Class']
train = train.drop(['Class'], axis=1)
return train, y
def get_stage2_test_data(preprocessed=False):
test = pd.read_csv('data/stage2_test_tag.csv')
return test
def train_test_split_by_gene(df, test_size=0.15, random_state=0):
np.random.seed(random_state)
genes = df.Gene.unique()
np.random.shuffle(genes)
index = np.zeros(len(df)).astype(bool)
thres = len(df) * (1 - test_size)
for gene in genes:
index[df['Gene'] == gene] = True
if np.sum(index) > thres:
break
test_index = np.logical_not(index)
return index, test_index
def get_amino_alias(amino, full=False):
data_file = 'data/one2many.json'
d = json.load(open(data_file))
if amino.upper() in d:
if not full:
return d[amino.upper()][-1]
else:
return d[amino.upper()][0]
return amino
def findall(text, idx=0):
alpha_list = set(list('abcdefghijklmnopqrstuvwxyz'))
words = text.lower().split()
found = []
for i, word in enumerate(words):
if not word[0] == '.':
continue
if not len(word)<20:
continue
if len(word) < 2:
continue
if len(word) == 2 and word[1] == 'a':
continue
if not word[1] in alpha_list:
continue
sent = ' '.join(words[i-4:i+5])
flag = False
for sp in ['.', ';', '(', ')', ':', '"', "'"]:
if sent.count(sp) > 3:
flag = True
break
if flag:
continue
flag = False
for sp in ['..', '--', ';;', '::', '""', "''"]:
if sp in sent:
flag = True
break
if flag:
continue
found.append(word)
return found
def get_fake_feature(df):
dot_count = df.apply(lambda row: len(findall(row['Text'], row['ID'])), axis=1)
var_count = df.apply(lambda row: row['Text'].lower().count(row['Variation'].lower()), axis=1)
var_freq = df.apply(lambda row: row['Text'].lower().count(row['Variation'].lower())/len(row['Text'].split()), axis=1)
gene_appear = df.apply(lambda row: row['Gene'].lower() in row['Text'].lower(), axis=1)
word_count = df.apply(lambda row: len(row['Text'].split()), axis=1)
return pd.concat([dot_count, var_count, var_freq, gene_appear, word_count], axis=1)
def write_submit_file(f, pred):
df_ans = pd.read_csv('data/sample_submission.csv')
real_id = np.array(pd.read_csv('data/stage2_test_real.csv')['ID'])
df_pred = pd.DataFrame(pred, columns=['class'+str(c+1) for c in range(9)])
df_pred['ID'] = real_id
for index, row in df_pred.iterrows():
df_ans.iloc[int(row['ID'])-1] = row
df_ans['ID'] = df_ans['ID'].astype(int)
df_ans.to_csv(f, index=False)