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#utility functions for neural net project
import random
def getNNPenData(fileString="datasets/pendigits.txt", limit=100000):
"""
returns limit # of examples from penDigits file
"""
examples=[]
data = open(fileString)
lineNum = 0
for line in data:
inVec = [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
outVec = [0,0,0,0,0,0,0,0,0,0] #which digit is output
count=0
for val in line.split(','):
if count==16:
outVec[int(val)] = 1
else:
inVec[count] = int(val)/100.0 #need to normalize values for inputs
count+=1
examples.append((inVec,outVec))
lineNum += 1
if (lineNum >= limit):
break
return examples
def getList(num,length):
list = [0]*length
list[num-1] = 1
return list
def getNNCarData(fileString ="datasets/car.data.txt", limit=100000 ):
"""
returns limit # of examples from file passed as string
"""
examples=[]
attrValues={}
data = open(fileString)
attrs = ['buying','maint','doors','persons','lug_boot','safety']
attr_values = [['vhigh', 'high', 'med', 'low'],
['vhigh', 'high', 'med', 'low'],
['2','3','4','5more'],
['2','4','more'],
['small', 'med', 'big'],
['high', 'med', 'low']]
attrNNList = [('buying', {'vhigh' : getList(1,4), 'high' : getList(2,4), 'med' : getList(3,4), 'low' : getList(4,4)}),
('maint',{'vhigh' : getList(1,4), 'high' : getList(2,4), 'med' : getList(3,4), 'low' : getList(4,4)}),
('doors',{'2' : getList(1,4), '3' : getList(2,4), '4' : getList(3,4), '5more' : getList(4,4)}),
('persons',{'2' : getList(1,3), '4' : getList(2,3), 'more' : getList(3,3)}),
('lug_boot',{'small' : getList(1,3),'med' : getList(2,3),'big' : getList(3,3)}),
('safety',{'high' : getList(1,3), 'med' : getList(2,3),'low' : getList(3,3)})]
classNNList = {'unacc' : [1,0,0,0], 'acc' : [0,1,0,0], 'good' : [0,0,1,0], 'vgood' : [0,0,0,1]}
for index in range(len(attrs)):
attrValues[attrs[index]]=attrNNList[index][1]
lineNum = 0
for line in data:
inVec = []
outVec = []
count=0
for val in line.split(','):
if count==6:
outVec = classNNList[val[:val.find('\n')]]
else:
inVec.append(attrValues[attrs[count]][val])
count+=1
examples.append((inVec,outVec))
lineNum += 1
if (lineNum >= limit):
break
return examples
def buildExamplesFromPenData(size=10000):
"""
build Neural-network friendly data struct
pen data format
16 input(attribute) values from 0 to 100
10 possible output values, corresponding to a digit from 0 to 9
"""
if (size != 10000):
penDataTrainList = getNNPenData("datasets/pendigitsTrain.txt",int(.8*size))
penDataTestList = getNNPenData("datasets/pendigitsTest.txt",int(.2*size))
else :
penDataTrainList = getNNPenData("datasets/pendigitsTrain.txt")
penDataTestList = getNNPenData("datasets/pendigitsTest.txt")
return penDataTrainList, penDataTestList
def buildExamplesFromCarData(size=200):
"""
build Neural-network friendly data struct
car data format
| names file (C4.5 format) for car evaluation domain
| class values - 4 value output vector
unacc, acc, good, vgood
| attributes
buying: vhigh, high, med, low.
maint: vhigh, high, med, low.
doors: 2, 3, 4, 5more.
persons: 2, 4, more.
lug_boot: small, med, big.
safety: low, med, high.
"""
carData = getNNCarData()
carDataTrainList = []
for cdRec in carData:
tmpInVec = []
for cdInRec in cdRec[0] :
for val in cdInRec :
tmpInVec.append(val)
#print "in :" + str(cdRec) + " in vec : " + str(tmpInVec)
tmpList = (tmpInVec, cdRec[1])
carDataTrainList.append(tmpList)
#print "car data list : " + str(carDataList)
tests = len(carDataTrainList)-size
carDataTestList = [carDataTrainList.pop(random.randint(0,tests+size-t-1)) for t in xrange(tests)]
return carDataTrainList, carDataTestList
def buildPotentialHiddenLayers(numIns, numOuts):
"""
This builds a list of lists of hidden layer layouts
numIns - number of inputs for data
some -suggestions- for hidden layers - no more than 2/3 # of input nodes per layer, and
no more than 2x number of input nodes total (so up to 3 layers of 2/3 # ins max
"""
resList = []
tmpList = []
maxNumNodes = max(numOuts+1, 2 * numIns)
if (maxNumNodes > 15):
maxNumNodes = 15
for lyr1cnt in range(numOuts,maxNumNodes):
for lyr2cnt in range(numOuts-1,lyr1cnt+1):
for lyr3cnt in range(numOuts-1,lyr2cnt+1):
if (lyr2cnt == numOuts-1):
lyr2cnt = 0
if (lyr3cnt == numOuts-1):
lyr3cnt = 0
tmpList.append(lyr1cnt)
tmpList.append(lyr2cnt)
tmpList.append(lyr3cnt)
resList.append(tmpList)
tmpList = []
return resList