물류 링크 기능이있는 SVM 회귀?

Question

저는 확률 론적 결과가 0과 1 사이 인 모델에 탄성 망 분류기를 사용하고 있습니다. SVM이 glmnet 모델보다 훨씬 더 우수한 분류 정확도를 제공한다는 것을 알았습니다. e1071이 SV 회귀를 지원한다는 것을 알고 있습니다. 링크 함수를 지정하고 벡터 로지스틱 회귀를 지원하는 방법이 있습니까? 감사. R 3.01, btw 사용.

Answer 1

로짓 링크 기능을 사용하는 대신 (나는 당신이 어떻게 될지 모르겠다), 모델 매개 변수에 probability=True을 설정해야한다고 생각한다.

이 예를 사용 해보세요 :

library(e1071) 

data(iris) 

#convert this into a two class problem with 0's and 1's in our response 
subs_iris = subset(iris, Species != 'virginica') 
subs_iris$Species = ifelse(subs_iris$Species == 'setosa',0,1) 
attach(subs_iris) 

x <- subset(subs_iris, select = -Species) 
y <- Species 
model <- svm(x, as.factor(y), probability=T) 
(pred <- predict(model, x, probability=T)) 


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 

43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 
0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 

64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 

85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 

attr(,"probabilities") 
       0   1 
1 0.987105973 0.01289403 
2 0.981947817 0.01805218 
3 0.985437173 0.01456283 
4 0.982502592 0.01749741 
5 0.986736251 0.01326375 
6 0.985797388 0.01420261 
7 0.982548458 0.01745154 
8 0.986618794 0.01338121 
9 0.978261097 0.02173890 
10 0.983057145 0.01694286 
11 0.985273718 0.01472628 
12 0.985044518 0.01495548 
13 0.982882845 0.01711715 
14 0.977350638 0.02264936 
15 0.977353717 0.02264628 
16 0.977337824 0.02266218 
17 0.986633849 0.01336615 
18 0.987559915 0.01244008 
19 0.979015418 0.02098458 
20 0.986641340 0.01335866 
21 0.979223478 0.02077652 
22 0.987128146 0.01287185 
23 0.977319879 0.02268012 
24 0.977318096 0.02268190 
25 0.982043578 0.01795642 
26 0.977302751 0.02269725 
27 0.985124984 0.01487502 
28 0.986267482 0.01373252 
29 0.985675766 0.01432423 
30 0.983434135 0.01656587 
31 0.982646443 0.01735356 
32 0.979790608 0.02020939 
33 0.979143578 0.02085642 
34 0.981984290 0.01801571 
35 0.983332565 0.01666743 
36 0.985862990 0.01413701 
37 0.980783392 0.01921661 
38 0.983998736 0.01600126 
39 0.979876054 0.02
40 0.986253654 0.01374635 
41 0.987705405 0.01229459 
42 0.946960057 0.05303994 
43 0.978967777 0.02103222 
44 0.980641733 0.01935827 
45 0.984315643 0.01568436 
46 0.981331513 0.01866849 
47 0.985423052 0.01457695 
48 0.983645068 0.01635493 
49 0.986290826 0.01370917 
50 0.986350301 0.01364970 
51 0.029433804 0.97056620 
52 0.016430412 0.98356959 
53 0.024776288 0.97522371 
54 0.009426853 0.99057315 
55 0.016452416 0.98354758 
56 0.012704127 0.98729587 
57 0.019445907 0.98055409 
58 0.030085332 0.96991467 
59 0.018511622 0.98148838 
60 0.014208017 0.98579198 
61 0.029383457 0.97061654 
62 0.014964845 0.98503516 
63 0.026215448 0.97378455 
64 0.013069314 0.98693069 
65 0.022539574 0.97746043 
66 0.019169611 0.98083039 
67 0.018805048 0.98119495 
68 0.019024824 0.98097518 
69 0.029443333 0.97055667 
70 0.011480062 0.98851994 
71 0.026777215 0.97322278 
72 0.014014692 0.98598531 
73 0.021568195 0.97843180 
74 0.015284531 0.98471547 
75 0.015807559 0.98419244 
76 0.016950698 0.98304930 
77 0.025482614 0.97451739 
78 0.023218517 0.97678148 
79 0.012583780 0.98741622 
80 0.022016637 0.97798336 
81 0.010875445 0.98912455 
82 0.014092065 0.98590794 
83 0.013192719 0.98680728 
84 0.017883555 0.98211644 
85 0.024212461 0.97578754 
86 0.029433804 0.97056620 
87 0.018419284 0.98158072 
88 0.025893165 0.97410684 
89 0.022054824 0.97794518 
90 0.008828270 0.99117173 
91 0.010761677 0.98923832 
92 0.013793894 0.98620611 
93 0.011470407 0.98852959 
94 0.023992095 0.97600791 
95 0.010806582 0.98919342 
96 0.022895843 0.97710416 
97 0.015274431 0.98472557 
98 0.014182305 0.98581770 
99 0.029373597 0.97062640 
100 0.012725531 0.98727447 
Levels: 0 1 

진실은, "단지 각 레코드 클래스에있는 확률을보고 내가 여기 $probabilities 속성을 액세스하는 어려움을 겪고있어,하지만 이론적으로 당신은 괜찮을 말했다 수를 1".