forked from amueller/kaggle_insults
-
Notifications
You must be signed in to change notification settings - Fork 0
/
results.txt
158 lines (113 loc) · 3.52 KB
/
results.txt
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
features:
uncommon words?,
quotes?
one-letter variations,
named entity,
any word twice?
ss50 0.899289015735
{'logr__C': 0.015625, 'vect__char_range': (1, 4), 'vect__word_range': (1, 2)}
0.896956114795
{'logr__C': 0.01, 'vect__char_range': (1, 5), 'vect__word_range': (1, 3)}
ss5 0.904839161565
{'logr__C': 0.015625, 'vect__char_range': (1, 5), 'vect__word_range': (1, 3)}
min_df=2
ss50 0.897950956364 +-0.01042787
{'logr__C': 0.01, 'vect__char_range': (1, 5), 'vect__word_range': (1, 3)}
min_df=1
ss50 0.899167675844 +-0.00996715
{'logr__C': 0.015625, 'vect__char_range': (1, 5), 'vect__word_range': (1, 2)}
min_df=1 nochar
0.889458659739
{'logr__C': 1.0, 'vect__word_range': (1, 2)}
min_df=2 nochar
0.889937560119
{'logr__C': 0.5, 'vect__word_range': (1, 3)}
only designed ss20:
0.754171776832
{'logr__C': 0.125}
just words ss20, mindf2
0.881163938504
{'logr__C': 0.25, 'vect__word_range': (1, 3)}
no words: ss10
0.906459334679
{'logr__C': 4.0, 'vect__char_range': (1, 4)}
no words: ss20
0.906488000861
{'logr__C': 4, 'vect__char_range': (1, 4)}
no words, more params, ss10
0.907882622299
{'logr__C': 6, 'vect__char_range': (1, 5)}
no words, more params, ss20
0.904469490972
{'logr__C': 9, 'vect__char_range': (1, 5)}
no words, classif selection
0.909000343987
{'logr__C': 9, 'select__percentile': 30, 'vect__word_range': (1, 3), 'vect__char_range': (1, 5)}
no words, chi2 selection
0.909025637541
{'logr__C': 16.0, 'select__percentile': 30, 'vect__word_range': (1, 3), 'vect__char_range': (1, 5)}
no words, chi2 selection
0.910659667085
{'logr__C': 13, 'select__percentile': 10, 'vect__word_range': (1, 3), 'vect__char_range': (1, 5)}
0.910878313902
{'logr__C': 20, 'select__percentile': 7, 'vect__word_range': (1, 3), 'vect__char_range': (1, 5)}
precompute features, ss50: coarse selction :-/ throw away
0.907881815965
{'logr__C': 16.0, 'select__percentile': 6}
0.911123747335
{'logr__C': 15, 'select__percentile': 8}
0.911813319384
{'logr__C': 15, 'select__percentile': 11}
ss 50:
0.910450058337
{'logr__C': 12, 'select__percentile': 9}
feature selection on words + chars, chi2:
0.912640914406
{'logr__C': 4.0, 'select__percentile': 16, 'vect__word_range': (1, 3), 'vect__char_range': (1, 5)}
0.913432113519
{'logr__C': 8, 'select__percentile': 15}
with updated badwords:
0.915528921865
{'logr__C': 14, 'select__percentile': 15
with counting bad words:
0.91333840187
{'logr__C': 9, 'select__percentile': 13}
ss20 binary=False, words + char
0.911491117421
{'logr__C': 6, 'select__percentile': 7}
0.913358142081
{'logr__C': 4.0, 'select__percentile': 15}
0.914403292468
{'logr__C': 5, 'select__percentile': 8}
ss 20, tfidf, binary=False, no scaling of designed features
0.912261756081
{'logr__C': 7, 'select__percentile': 2}
with scaling of features:
0.912951813566
{'logr__C': 3, 'select__percentile': 10}
l1 select: ss5
char+words+designed, (1, 5), (1, 3)
0.897332385116
{'logr__C': 4.0, 'select__C': 8.0}
no select ss50:
0.903399470341
{'logr__C': 5}
min_df ss10
0.917669556837
{'logr__C': 2, 'select__percentile': 14, 'vect__words__min_df': 3}
ss10 new feature extraction. only words
0.893394841981
{'C': 4}
ne features, only words and stuff
0.903229621733
{'select__percentile': 16}
elasticnet: no select
0.906497374319
{'logr__alpha': 0.0001220703125, 'logr__rho': 0.8, 'logr__n_iter': 5}
0.912700394265
{'logr__alpha': 0.0001, 'logr__rho': 1, 'logr__n_iter': 11}
0.912545571475
{'logr__alpha': 0.0001, 'logr__rho': 0.94999999999999996, 'logr__n_iter': 21}
select percentile
0.914085740028
{'logr__alpha': 6.103515625e-05, 'select__percentile': 11}