> head(iris)
Sepal.Length Sepal.Width Petal.Length Petal.Width Species
1 5.1 3.5 1.4 0.2 setosa
2 4.9 3.0 1.4 0.2 setosa
3 4.7 3.2 1.3 0.2 setosa
4 4.6 3.1 1.5 0.2 setosa
5 5.0 3.6 1.4 0.2 setosa
6 5.4 3.9 1.7 0.4 setosa
> cor(iris[1:4]) #상관계수 확인
Sepal.Length Sepal.Width Petal.Length Petal.Width
Sepal.Length 1.0000000 -0.1175698 0.8717538 0.8179411
Sepal.Width -0.1175698 1.0000000 -0.4284401 -0.3661259
Petal.Length 0.8717538 -0.4284401 1.0000000 0.9628654
Petal.Width 0.8179411 -0.3661259 0.9628654 1.0000000
> log.ir<-log(iris[,1:4]) #데이터가 편향되어 있으므로 log변환 해주기
> ir.species<-iris[,5] #species는 y변수
> #주성분 분석
> ir.pca<-prcomp(log.ir, center=T, scale.=T)
> print(ir.pca)
Standard deviations (1, .., p=4):
[1] 1.7124583 0.9523797 0.3647029 0.1656840
Rotation (n x k) = (4 x 4):
PC1 PC2 PC3 PC4
Sepal.Length 0.5038236 -0.45499872 0.7088547 0.19147575
Sepal.Width -0.3023682 -0.88914419 -0.3311628 -0.09125405
Petal.Length 0.5767881 -0.03378802 -0.2192793 -0.78618732
Petal.Width 0.5674952 -0.03545628 -0.5829003 0.58044745
> plot(ir.pca, type="l")
> 가 급 꺾이는데 이 부분을 elbow point라고 하며,
Error: unexpected symbol in "가 급"
> #이 위의 주성분을 주로 선택하여 사용함
>
> summary(ir.pca)
Importance of components:
PC1 PC2 PC3 PC4
Standard deviation 1.7125 0.9524 0.36470 0.16568
Proportion of Variance 0.7331 0.2268 0.03325 0.00686
Cumulative Proportion 0.7331 0.9599 0.99314 1.00000
> #PC1의 Cumulative~의 데이터를 보면 PC1이 73% 설명 가능하고
> #PC2까지 추가되면 95%까지 설명가능한 것
>
> #원래 데이터와 선형계수를 메트릭스 곱으로 선형조합에 맞게 만들기
> #ir.pca$rotation에 선혀계수 있음
> PRC<-as.matrix(log.ir) %*% ir.pca$rotation
> head(PRC)
PC1 PC2 PC3 PC4
[1,] -0.2772209 -1.809493 1.604387 -1.0010840
[2,] -0.2507663 -1.654229 1.627078 -0.9946772
[3,] -0.3340210 -1.690148 1.592416 -0.9502831
[4,] -0.2527176 -1.656968 1.556306 -1.0640079
[5,] -0.2957159 -1.825531 1.581020 -1.0074464
[6,] 0.2242011 -1.962854 1.162457 -0.7503219
>
> train1 <- cbind(ir.species,as.data.frame(PRC))
> train1[,1] <- as.factor(train1[,1])
> colnames(train1)[1] <- "label"
>
> head(train1)
label PC1 PC2 PC3 PC4
1 setosa -0.2772209 -1.809493 1.604387 -1.0010840
2 setosa -0.2507663 -1.654229 1.627078 -0.9946772
3 setosa -0.3340210 -1.690148 1.592416 -0.9502831
4 setosa -0.2527176 -1.656968 1.556306 -1.0640079
5 setosa -0.2957159 -1.825531 1.581020 -1.0074464
6 setosa 0.2242011 -1.962854 1.162457 -0.7503219
>
> #회귀분석
> fit1 <- lm(label~PC1+PC2, data=train1)
Warning messages:
1: In model.response(mf, "numeric") :
요인형 종속변수의 type = "numeric"의 사용은 무시될 것입니다
2: In Ops.factor(y, z$residuals) : ‘-’ not meaningful for factors
> fit1
Call:
lm(formula = label ~ PC1 + PC2, data = train1)
Coefficients:
(Intercept) PC1 PC2
0.6696 0.7680 -0.2548
> #이 회귀모델의 예측력이 좋은지 확인하기 위해 predict함수로 예측해 봄
> fit1_pred <- predict(fit1,newdata = train1)
> fit1_pred
1 2 3 4 5 6 7
0.9177087 0.8984698 0.8436802 0.8976691 0.9075902 1.3418786 1.0463684
8 9 10 11 12 13 14
0.9390723 0.8445396 0.6210744 0.9772913 0.9476893 0.5797361 0.4155148
15 16 17 18 19 20 21
0.9120119 1.3118574 1.2207336 1.0980907 1.2412319 1.1287785 1.0342953
22 23 24 25 26 27 28
1.2569135 0.7137137 1.4133590 1.0252950 0.9689303 1.2765822 0.9586297
29 30 31 32 33 34 35
0.9276385 0.9374478 0.9482156 1.2861381 0.6493632 0.9546398 0.9294394
36 37 38 39 40 41 42
0.8386486 0.9222122 0.5890660 0.8108801 0.9490303 1.0546664 1.0040768
43 44 45 46 47 48 49
0.8105124 1.4567990 1.3635121 1.0684832 0.9775414 0.8663319 0.9678917
50 51 52 53 54 55 56
0.9080860 2.4900652 2.4560586 2.5325228 2.2648798 2.4745413 2.3349098
57 58 59 60 61 62 63
2.4963131 2.0029576 2.4183566 2.2572967 2.0407272 2.3843643 2.1921680
64 65 66 67 68 69 70
2.4214215 2.2250403 2.4384336 2.3892785 2.1851043 2.4422283 2.1877138
71 72 73 74 75 76 77
2.5254085 2.3158257 2.4880022 2.3530434 2.3724269 2.4310584 2.4857568
78 79 80 81 82 83 84
2.5827240 2.4241655 2.1051215 2.1671554 2.1127110 2.2436308 2.5098066
85 86 87 88 89 90 91
2.3709906 2.4519709 2.4989128 2.3762102 2.2835777 2.2644047 2.2716131
92 93 94 95 96 97 98
2.4115164 2.2552790 2.0133592 2.2950601 2.2677511 2.3035534 2.3564617
99 100 101 102 103 104 105
2.0222024 2.2928712 2.8051324 2.5692110 2.7806345 2.6285684 2.7492116
106 107 108 109 110 111 112
2.8654869 2.3788516 2.7558419 2.6762160 2.8792490 2.6483604 2.6360837
113 114 115 116 117 118 119
2.7272214 2.5747804 2.6729336 2.7201117 2.6359540 2.8982001 2.9334209
120 121 122 123 124 125 126
2.4733191 2.7907960 2.5561109 2.8575388 2.5686743 2.7353583 2.7263118
127 128 129 130 131 132 133
2.5511097 2.5518512 2.7052657 2.6589710 2.7723682 2.8480064 2.7259613
134 135 136 137 138 139 140
2.5054224 2.5011639 2.8769550 2.7556452 2.6279707 2.5342278 2.7260895
141 142 143 144 145 146 147
2.7871266 2.7407486 2.5692110 2.7990284 2.8129241 2.7349133 2.6022894
148 149 150
2.6574971 2.7122422 2.5531536
>
> b <- round(fit1_pred) #반올림 함수
>
> b[b==0 | b==1] <- "setosa"
> b[b==2] <- "Versicolor"
> b[b==3] <- "Virginica"
>
> a <- ir.species
> table(b,a)
a
b setosa versicolor virginica
setosa 50 0 0
Versicolor 0 46 2
Virginica 0 4 48
>
> #setosa : 50단위 中 50단위 예측
> #versicolor : 50단위 中 46단위 예측
> #virginica : 50단위 中 48단위 예측
출처 : https://m.blog.naver.com/PostView.nhn?blogId=leedk1110&logNo=220783514855&proxyReferer=https%3A%2F%2Fwww.google.co.kr%2F
'programing > R studio' 카테고리의 다른 글
| 상관계수 - cor(), 다중공선성 (0) | 2018.05.08 |
|---|---|
| attach()와 detach() (0) | 2018.05.08 |
| 네트워크 분석 개요 (0) | 2018.05.08 |
| 네트워크 분석 - igraph (0) | 2018.05.08 |
| facet_wrap() (0) | 2018.05.07 |