-
Notifications
You must be signed in to change notification settings - Fork 1
/
titanicsvmOLD.py
201 lines (166 loc) · 5.85 KB
/
titanicsvmOLD.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
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
#Import whatever
import csv as csv
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.svm import LinearSVC
from sklearn.svm import SVC
from sklearn.cross_validation import StratifiedKFold
from sklearn.grid_search import GridSearchCV
from collections import Counter
from pandas import *
def cabin_letter(x):
return{
'A':1,
'B':2,
'C':3,
'D':4,
'E':5,
'F':6,
'G':7,
}.get(x,0)
def num (s):
try:
return int(s)
except ValueError:
return float(s)
def fixdataSVM (data):
#split the cabin column to find the last split (the last if there are multiple cabins)
#then assign a number based on the cabin letter
cabin_letter_list = []
for i in data[0::,9]:
splitter = i.rsplit(' ',1)[-1]
if splitter:
splitter = cabin_letter(splitter[0])
else:
splitter = 0
cabin_letter_list.append(splitter)
data[0::,9] = cabin_letter_list
#print data[0,9]
#Male = 1, female = 0:
data[data[0::,3]=='male',3] = 1
data[data[0::,3]=='female',3] = 0
#embark c=0, s=1, q=2
data[data[0::,10] =='C',10] = 0
data[data[0::,10] =='S',10] = 1
data[data[0::,10] =='Q',10] = 2
data[data[0::,10] == '',10] = 3
#find the most common embark point, put it in the blanks
data[data[0::,10] == 3,10] = max(Counter(data[data[0::,10] != 3,10]))
#take means for age for each class
age = []
firstclassfare = []
secondclassfare = []
thirdclassfare = []
for x in data:
if x[4]:
age.append(np.float(x[4]))
if x[1] == '1' and x[8]:
firstclassfare.append(np.float(x[8]))
if x[1] == '2' and x[8]:
secondclassfare.append(np.float(x[8]))
if x[1] == '3' and x[8]:
thirdclassfare.append(np.float(x[8]))
ageaverage = int(np.mean(age))
firstclassfareaverage = int(np.mean(firstclassfare))
secondclassfareaverage = int(np.mean(secondclassfare))
thirdclassfareaverage = int(np.mean(thirdclassfare))
#put those averages back into the '' ages
for i in xrange(np.size(data[0::,0])):
try:
float(data[i,4])
except ValueError:
data[i,4] = ageaverage
try:
float(data[i,8])
except ValueError:
if data[i,1] == '1':
data[i,8] = firstclassfareaverage
if data[i,1] == '2':
data[i,8] = secondclassfareaverage
if data[i,1] == '3':
data[i,8] = thirdclassfareaverage
#clean up the name and ticket elements
data = np.delete(data,[2,7],1)
#change strings to float
for i in xrange(np.size(data[0::,0])):
for y in range(9):
try:
data[i,y] = num(data[i,y])
except ValueError:
print y
print data[i,y]
print '--'
#data[i,y] = num(0)
return data
def scaleData (trainer,tester):
deparray = trainer[0::,0]
df = DataFrame(trainer.astype(np.float))
df_norm = (df-df.mean())/(df.max()-df.min())
trainer = np.array(df_norm)
trainer[0::,0] = deparray
dfT = DataFrame(tester.astype(np.float))
dfT_scaled = (dfT-df.mean())/(df.max()-df.min())
tester = np.array(dfT_scaled)
tester[0::,0] = 0
bothArray = [trainer,tester]
return bothArray
#import training data
csv_file_object = csv.reader(open('train.csv', 'rb')) #Load in the csv file
header = csv_file_object.next() #Skip the fist line as it is a header
train_data=[] #Creat a variable called 'data'
for row in csv_file_object: #Skip through each row in the csv file
train_data.append(row) #adding each row to the data variable
train_data = np.array(train_data) #Then convert from a list to an array
test_file_object = csv.reader(open('test.csv', 'rb')) #Load in the test csv file
header = test_file_object.next() #Skip the fist line as it is a header
test_data=[] #Creat a variable called 'test_data'
for row in test_file_object: #Skip through each row in the csv file
test_data.append(row) #adding each row to the data variable
test_data = np.array(test_data) #Then convert from a list to an array
test_data = np.insert(test_data,[0], 0, axis=1)
#normalize data frame, remove ticket and name, fix cabin to be just the letter
#in the future, scale the data by the train set and apply that to test set
train_data = fixdataSVM(train_data)
test_data = fixdataSVM(test_data)
train_test = scaleData(train_data, test_data)
train_data = train_test[0]
test_data = train_test[1]
#print train_data[2]
#print test_data[2]
#do a quick forest
forest = RandomForestClassifier(n_estimators=100)
forest = forest.fit(train_data[0::,1::],train_data[0::,0])
forest_output = forest.predict(test_data[0::,1::])
open_file_object = csv.writer(open("normalizedforest.csv", "wb"))
test_file_object = csv.reader(open('test.csv', 'rb')) #Load in the csv file
test_file_object.next()
i = 0
for row in test_file_object:
row.insert(0,forest_output[i].astype(np.uint8))
open_file_object.writerow(row)
i += 1
#do a grid search for c,y on the data, possibly a second better-region-only search
#do a bigger fold, maybe 10 instead of 3
C_range = 2.0 ** np.arange(-13,30)
gamma_range = 2.0 ** np.arange(-17,13)
param_grid = dict(gamma=gamma_range, C=C_range)
cv = StratifiedKFold(train_data[0::,0], 10)
grid = GridSearchCV(SVC(kernel='rbf',cache_size=2000), param_grid = param_grid, cv = cv, n_jobs=5, verbose=2)
grid.fit(train_data[0::,1::],train_data[0::,0])
thebest = ["The best classifier is: ", grid.best_estimator_]
open_file_object = csv.writer(open("thebest.txt", "wb"))
open_file_object.writerow(thebest)
open_file_object.writerow(train_data[5])
open_file_object.writerow(test_data[15])
#run an RBF kernel on the data
myRBF = grid.best_estimator_
rbfResults = myRBF.predict(test_data[0::,1::])
#record the RBF results
open_file_object = csv.writer(open("rbf.csv", "wb"))
test_file_object = csv.reader(open('test.csv', 'rb')) #Load in the csv file
test_file_object.next()
i = 0
for row in test_file_object:
row.insert(0,rbfResults[i].astype(np.uint8))
open_file_object.writerow(row)
i += 1