R言語で統計分類基本

この発表はR言語を使って判別分析を実装するほうほうを紹介する

Transcript

1. 関東第4回ゼロからはじめる Ｒ言語勉強会 R言語で統計分類基本 パヴェウ・ルシン 　株式会社ブリリアントサービス
2. 自己紹介 •Paweł Rusin (パヴェウ・ルシン) •pawel.rusin@brilliantservice.co.jp Facebook: Paweł Rusin ( iofar@o2.pl) •会社： •　株式会社ブリリアントサービス •業務： データマイニング
3. 分類 • 統計分類というのは個体をグループ分けする統計の手続きです • 事前にラベル付けされた訓練例を使ってははじめて見たオブジェクト 分類できるようになります 学習データ テストデータ 機会学習 分類
4. 分類の例 spam not spam spam not spam 件名：【重要】５分以内に必ず確認 ※退会を希望する場合には↓へ 件名：ご協力求む！ 友人の竹山さんからのメールを転送しました！ いつも大変お世話になっております。 メール部品販売株式会社　営業２課　営業勇作です。 spam 件名：おひさしぶりです＾＾ 覚えてますか？？？ ? 訓練例： テスト例：
5. 分類の例 pregnant glucose pressure triceps insulin mass pedigree age diabetes 1 6 148 72 35 NA 33.6 0.627 50 pos 2 1 85 66 29 NA 26.6 0.351 31 neg 3 8 183 64 NA NA 23.3 0.672 32 pos 4 1 89 66 23 94 28.1 0.167 21 neg 5 0 137 40 35 168 43.1 2.288 33 pos 6 5 116 74 NA NA 25.6 0.201 30 neg pregnant glucose pressure triceps insulin mass pedigree age diabetes 763 9 89 62 NA NA 22.5 0.142 33 ? 764 10 101 76 48 180 32.9 0.171 63 ? 765 2 122 70 27 NA 36.8 0.340 27 ? 766 5 121 72 23 112 26.2 0.245 30 ? 767 1 126 60 NA NA 30.1 0.349 47 ? 768 1 93 70 31 NA 30.4 0.315 23 ? 訓練例： テスト例：
6. 分類の例 WHO Risk Group 1 WHO Risk Group 2 WHO Risk Group 3 WHO Risk Group 4 time status sex age year thickness ulcer 1 10 3 1 76 1972 6.76 1 2 30 3 1 56 1968 0.65 0 3 35 2 1 41 1977 1.34 0 4 99 3 0 71 1968 2.90 0 5 185 1 1 52 1965 12.08 1 6 204 1 1 28 1971 4.84 1 7 210 1 1 77 1972 5.16 1 8 232 3 0 60 1974 3.22 1 9 232 1 1 49 1968 12.88 1 10 279 1 0 68 1971 7.41 1
7. データを片付ける pregnant glucose pressure triceps insulin mass pedigree age diabetes 1 6 148 72 35 NA 33.6 0.627 50 pos 2 1 85 66 29 NA 26.6 0.351 31 neg 3 8 183 64 NA NA 23.3 0.672 32 pos 4 1 89 66 23 94 28.1 0.167 21 neg 5 0 137 40 35 168 43.1 2.288 33 pos 6 5 116 74 NA NA 25.6 0.201 30 neg > install.packages("MASS") > library(MASS) > data(PimaIndiansDiabetes2) > indians = na.omit(PimaIndiansDiabetes2) > indians = indians[,c(2,5,9)] glucose insulin diabetes 4 89 94 neg 5 137 168 pos 7 78 88 pos 9 197 543 pos 14 189 846 pos 15 166 175 pos
8. 訓練例とテスト例を分ける learning.sample = sample(x=1:nrow(indians),size=nrow(indians)/2) learning.set = indians[learning.sample,] sample(x, size, replace = FALSE, prob = NULL) x=[1,2...392],size=196 [1] 120 321 292 11 49 318 ... glucose insulin diabetes 317 89 94 neg 318 137 168 pos 319 78 88 pos 320 197 543 pos 321 189 846 pos 322 166 175 pos [[120,321,292,11,49,318...],]
9. 訓練例とテスト例を分ける test.set = indians[-learning.set,-3] learning.set = sample(x=[1,2...392],size=196) [1] 120 321 292 11 49 318 ... glucose insulin diabetes 317 89 94 neg 318 137 168 pos 319 78 88 pos 320 197 543 pos 321 189 846 pos 322 166 175 pos [-[120,321,292,11,49,318...],-3]
10. 線形判別分析
11. 線形判別分析
12. 線形判別分析
13. 線形判別分析 lin.classify = lda(indians[,1:2],grouping=indians$diabetes,subset=learning.sample) lda(x, grouping, ..., subset, na.action) [MASS] Call: lda(diabetes ~ glucose + insulin, data = indians.formatted, subset = learning.set.sample) Prior probabilities of groups: neg pos 0.6377551 0.3622449 Group means: glucose insulin neg 113.2400 135.1520 pos 146.4366 208.0282 Coefficients of linear discriminants: LD1 glucose 0.0353017338 insulin 0.0008873364
14. 線形判別分析 lin.class.predict = predict(lin.classify, newdata=test.set) predict (object, ...) $class [1] neg neg pos neg neg pos neg neg neg pos neg neg neg neg neg neg neg neg neg neg neg neg neg neg neg pos neg neg neg pos neg neg neg neg [35] pos neg pos neg neg neg neg pos neg neg pos neg pos neg neg neg neg neg pos neg neg neg pos neg neg neg neg neg pos neg neg neg neg neg [69] neg neg neg neg neg pos pos neg pos neg neg neg neg neg pos pos neg neg pos neg neg neg pos neg neg neg neg neg neg neg neg neg pos neg [103] pos neg neg pos pos neg neg neg neg neg neg neg neg pos neg neg neg neg neg neg neg neg pos neg neg neg pos neg neg pos neg neg neg neg [137] neg pos neg neg neg neg neg neg neg neg pos neg neg neg neg neg pos pos pos pos neg neg neg neg neg neg neg neg pos pos neg neg neg neg [171] pos pos pos neg neg neg neg neg neg neg pos pos neg neg neg neg pos neg neg neg pos neg neg neg neg neg Levels: neg pos $posterior neg pos 4 0.9100464 0.08995355 5 0.5700686 0.42993137 14 0.2765809 0.72341915 21 0.7073037 0.29269633 ... res.table = table(real=true.test.set,classified=lin.class.predict$class) classified real neg pos neg 102 28 pos 51 15
15. 線形判別分析 drawparti(grouping,x,y,method=”lda”...) [klaR] drawparti(indians[,3],indians[,1],indians[,2]...) drawparti(grouping, x, y, method = "lda", prec = 100, xlab = NULL, ylab = NULL, col.correct = "black", col.wrong = "red", col.mean = "black", col.contour = "darkgrey", gs = as.character(grouping), pch.mean = 19, cex.mean = 1.3, print.err = 0.7, legend.err = FALSE, legend.bg = "white", imageplot = TRUE, image.colors = cm.colors(nc), plot.control = list(), ...)
16. 二次判別分析
17. quad.classify = qda(indians[,1:2],grouping=indians$diabetes,subset=learning.sample) qda(x, grouping, ..., subset, na.action) [MASS] Call: qda(indians.formatted[, 1:2], grouping = indians.formatted$diabetes, subset = learning.set.sample) Prior probabilities of groups: neg pos 0.6377551 0.3622449 Group means: glucose insulin neg 113.2400 135.1520 pos 146.4366 208.0282 二次判別分析
18. quad.class.predict = predict(quad.classify, newdata=test.set) predict (object, ...) res.table = table(real=true.test.set,classified=quad.class.predict$class) $class [1] neg neg pos neg neg pos neg neg neg pos neg neg neg neg neg neg neg neg neg neg neg neg neg neg neg pos neg neg neg pos neg neg neg neg [35] pos neg pos neg neg neg neg pos neg neg pos neg pos neg neg neg neg neg pos neg neg neg pos neg neg neg pos neg pos neg neg neg neg pos [69] neg neg neg neg neg pos neg neg pos neg neg neg neg neg pos pos neg neg pos neg neg neg pos neg neg neg neg neg neg neg neg neg pos neg [103] pos neg neg pos pos neg neg neg neg neg neg neg neg neg neg neg neg neg neg neg neg neg pos neg neg neg pos neg neg pos neg neg neg neg [137] neg pos neg neg neg neg neg neg neg neg pos neg neg neg neg pos pos pos pos pos neg neg neg neg neg neg neg neg pos pos neg neg neg neg [171] pos pos pos neg neg neg neg neg neg neg pos pos neg neg neg neg pos neg neg neg pos neg neg neg neg neg Levels: neg pos $posterior neg pos 4 0.903509648 0.09649035 5 0.651253333 0.34874667 14 0.000552298 0.99944770 classified real neg pos neg 103 27 pos 49 17 二次判別分析
19. drawparti(indians[,3],indians[,1],indians[,2],method=”qda”...) 二次判別分析
20. 分類器 R言語のパッケージと関数 線形判別分析,二次判別分析 MASS(lda,qda) 単純バイズ分類器 e1071(naiveBayes), klaR(NaiveBayes) 決定木 tree(tree),rpart(rpart),party(cpart) Random Forest randomForest(randomForest) k近傍法 class(knn),kknn(kknn),knncat(knncat) サポートバクターマシン e1071(svm) ニューラルネットァーク nnet(nnet) 意外の分類器
21. 意外の分類器 データを処理: • 属性の時限を減る • NAの値を扱う(na.omitとか。。。) • 学習セットを選ぶ(sampleとか。。。) 学習 Lda()とかqda()とかnaiveBayes()など。。。 分類 predict() データフレーム 分類器のオブジェクト
22. R言語勉強会を参加していただいて ありがとうございました！ Facebook: Paweł Rusin (iofar@o2.pl)