決定木Decision Tree：rpart.control( )
データの作成と取得
「kguの日記」さんをなぞっていきます。
まずはサンプルデータを手作業で作成し、それをread.table( )でRに取り込みます。
gender age income most_cost
male 22 330 food
male 25 390 food
male 29 450 food
male 31 490 beauty
male 35 520 learning
male 39 640 car
male 40 630 car
male 47 600 car
male 49 790 car
male 51 530 learning
male 53 540 learning
male 57 800 car
male 60 450 food
male 68 330 food
male 72 310 food
female 22 310 food
female 24 320 food
female 27 400 food
female 31 410 food
female 35 450 beauty
female 37 530 beauty
female 42 590 beauty
female 47 700 car
female 49 600 beauty
female 50 420 food
female 55 510 beauty
female 59 390 food
female 62 600 learning
female 63 350 food
female 79 300 food
data <- read.table("/Users/kazuki/Desktop/data.txt",header=T)
決定木decision treeの作成
rpart, partykitライブラリを用いて決定木のグラフを書きます。
> library(rpart)
> library(partykit)
rpctrl <- rpart.control(minplit=10)
> model = rpart(most_cost~., data, control=rpctrl)
> plot(as.party(model))
決定木による予測
テストデータを作成します。
gender age income most_cost
male 45 290 food(exp)
male 57 730 car(exp)
female 33 490 beauty(exp)
このデータから、先ほどの決定木にかけてみることで予測が可能です。(要考！)
> test <- read.table("/Users/kazuki/Desktop/test.txt",header = T)
> pred <- predict(model, test)
> pred
beauty car food learning
1 0.06666667 0.0 0.9333333 0.0000000
2 0.33333333 0.4 0.0000000 0.2666667
3 0.33333333 0.4 0.0000000 0.2666667