forked from jingxil/Neural-Decision-Forests
-
Notifications
You must be signed in to change notification settings - Fork 0
/
dataset.py
171 lines (137 loc) · 8.52 KB
/
dataset.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
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
import torch
from torch.utils.data import Dataset
import os
import numpy as np
from sklearn.model_selection import train_test_split
class UCIAdult(Dataset):
def __init__(self,root,train=True):
super(UCIAdult,self).__init__()
self.root = os.path.expanduser(root)
self.train = train
# 65, Private, 198766, Masters, 14, Married-civ-spouse, Sales, Husband, White, Male, 20051, 0, 40, United-States, >50K
self.fields = [{'type':'num'},
{'type':'cate','choices':['unk','Private', 'Self-emp-not-inc', 'Self-emp-inc', 'Federal-gov', 'Local-gov', 'State-gov', 'Without-pay', 'Never-worked']},
{'type': 'num'},
{'type': 'cate', 'choices': ['unk','Bachelors', 'Some-college', '11th', 'HS-grad', 'Prof-school', 'Assoc-acdm', 'Assoc-voc', '9th', '7th-8th', '12th', 'Masters', '1st-4th', '10th', 'Doctorate', '5th-6th', 'Preschool']},
{'type': 'num'},
{'type': 'cate', 'choices': ['unk','Married-civ-spouse', 'Divorced', 'Never-married', 'Separated', 'Widowed', 'Married-spouse-absent', 'Married-AF-spouse' ]},
{'type': 'cate', 'choices': ['unk','Tech-support', 'Craft-repair', 'Other-service', 'Sales', 'Exec-managerial', 'Prof-specialty', 'Handlers-cleaners', 'Machine-op-inspct', 'Adm-clerical', 'Farming-fishing', 'Transport-moving', 'Priv-house-serv', 'Protective-serv', 'Armed-Forces' ]},
{'type': 'cate', 'choices': ['unk', 'Wife', 'Own-child', 'Husband', 'Not-in-family', 'Other-relative', 'Unmarried']},
{'type': 'cate', 'choices': ['unk', 'White', 'Asian-Pac-Islander', 'Amer-Indian-Eskimo', 'Other', 'Black']},
{'type': 'cate', 'choices': ['unk', 'Female', 'Male']},
{'type': 'num'},
{'type': 'num'},
{'type': 'num'},
{'type': 'cate', 'choices': ['unk', 'United-States', 'Cambodia', 'England', 'Puerto-Rico', 'Canada', 'Germany', 'Outlying-US(Guam-USVI-etc)', 'India', 'Japan', 'Greece', 'South', 'China', 'Cuba', 'Iran', 'Honduras', 'Philippines', 'Italy', 'Poland', 'Jamaica', 'Vietnam', 'Mexico', 'Portugal', 'Ireland', 'France', 'Dominican-Republic', 'Laos', 'Ecuador', 'Taiwan', 'Haiti', 'Columbia', 'Hungary', 'Guatemala', 'Nicaragua', 'Scotland', 'Thailand', 'Yugoslavia', 'El-Salvador', 'Trinadad&Tobago', 'Peru', 'Hong', 'Holand-Netherlands']},
]
for field in self.fields:
if field['type'] == 'cate':
field['w2i'] = dict([(w,i) for i,w in enumerate(field['choices'])])
self.X, self.y = self.load_data(root,train)
def __getitem__(self, index):
return (self.X[index],self.y[index])
def __len__(self):
return len(self.X)
def load_data(self,root,train):
if train:
data_path = os.path.join(root,"adult.data")
else:
data_path = os.path.join(root, "adult.test")
with open(data_path) as f:
rows = [ [ fv.strip() for fv in row.strip().split(",")] for row in f.readlines() if len(row.strip()) > 0 and not row.startswith("|")]
n_datas = len(rows)
X_dim = np.sum([ 1 if field['type']=='num' else len(field['choices']) for field in self.fields])
X = np.zeros((n_datas, X_dim), dtype=np.float32)
y = np.zeros(n_datas, dtype=np.int32)
for i, row in enumerate(rows):
assert len(row) == 15
foffset = 0
for j in range(14):
if self.fields[j]['type'] == 'num':
fdim=1
X[i, foffset] = float(row[j])
else:
fdim = len(self.fields[j]['choices'])
hit = self.fields[j]['w2i'].get(row[j],0)
X[i, foffset+hit] = 1
foffset += fdim
y[i] = 0 if row[-1].strip().startswith("<=50K") else 1
X_min = np.array([[17.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 12285.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]])
X_max = np.array([[90.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1484705.0, 0.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 16.0, 0.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 0.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 0.0, 1.0, 1.0, 1.0, 1.0, 1.0, 0.0, 1.0, 1.0, 99999.0, 4356.0, 99.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]])
X = (X - X_min) / np.clip((X_max - X_min), a_min=1e-6, a_max=None)
#print([float(i) for i in np.min(X, axis=0)])
#print([float(i) for i in np.max(X, axis=0)])
X = torch.from_numpy(X).type(torch.FloatTensor)
y = torch.from_numpy(y).type(torch.IntTensor)
return X, y
class UCILetter(Dataset):
def __init__(self,root,train=True):
super(UCILetter,self).__init__()
self.root = os.path.expanduser(root)
self.train = train
self.X, self.y = self.load_data(root,train)
def __getitem__(self, index):
return (self.X[index],self.y[index])
def __len__(self):
return len(self.X)
def load_data(self,root,train):
data_path = os.path.join(root, "letter-recognition.data")
with open(data_path) as f:
rows = [[ item.strip() for item in row.strip().split(',')] for row in f.readlines()]
if train:
rows = rows[:16000]
else:
rows = rows[16000:]
n_datas = len(rows)
X = np.zeros((n_datas, 16), dtype=np.float32)
y = np.zeros(n_datas, dtype=np.int32)
for i, row in enumerate(rows):
X[i, :] = list(map(float, row[1:]))
y[i] = ord(row[0]) - ord('A')
X = torch.from_numpy(X).type(torch.FloatTensor)
y = torch.from_numpy(y).type(torch.IntTensor)
return X,y
class UCIYeast(Dataset):
def __init__(self,root,train=True):
super(UCIYeast,self).__init__()
self.root = os.path.expanduser(root)
self.train = train
self.labels = ['CYT','NUC','MIT','ME3','ME2','ME1','EXC','VAC','POX','ERL']
self.labels2idx = dict([ (label,i) for i,label in enumerate(self.labels)])
self.X, self.y = self.load_data(root,train)
def __getitem__(self, index):
return (self.X[index],self.y[index])
def __len__(self):
return len(self.X)
def _load_data(self,data_path):
with open(data_path) as f:
rows = [ row.strip().split() for row in f.readlines()]
n_datas = len(rows)
X = np.zeros((n_datas, 8), dtype=np.float32)
y = np.zeros(n_datas, dtype=np.int32)
for i, row in enumerate(rows):
X[i, :] = list(map(float, row[1:-1]))
y[i] = self.labels2idx[row[-1]]
return X,y
def _write_data(self, X, y, data_path):
with open(data_path,'w') as f:
for X_, y_ in zip(X,y):
feat = ' '.join([str(x) for x in X_])
f.write('placeholder %s %s\n'%(feat,self.labels[y_]))
def load_data(self,root,train):
if not os.path.exists(os.path.join(root, "yeast.train")) or \
not os.path.exists(os.path.join(root, "yeast.test")):
data_path = os.path.join(root, "yeast.data")
X,y = self._load_data(data_path)
n_datas = len(y)
train_idx, test_idx = train_test_split(range(n_datas), random_state=0, train_size=0.7, stratify=y)
self._write_data(X[train_idx],y[train_idx],os.path.join(root, "yeast.train"))
self._write_data(X[test_idx], y[test_idx], os.path.join(root, "yeast.test"))
if train:
data_path = os.path.join(root, "yeast.train")
else:
data_path = os.path.join(root, "yeast.test")
X, y = self._load_data(data_path)
X = torch.from_numpy(X).type(torch.FloatTensor)
y = torch.from_numpy(y).type(torch.IntTensor)
return X,y