2.1 Regresi Linier
Program 2.1
data <- read.csv("data sapi.csv")
data <- data[,-1]
regresi <- lm(Ybbt_badan ~ ., data=data)
regresi
Output 2.1
##
## Call:
## lm(formula = Ybbt_badan ~ ., data = data)
##
## Coefficients:
## (Intercept) X1pjg_badan X2ttg_badan X3lkr_badan X4pjg_ekor
## 118.8305 6.1877 4.5001 7.4794 1.1279
## X5pjg_tanduk
## 0.1448
Program 2.2
data <- read.csv("data sapi.csv")
data <- data[,-1]
ambil.uji <- sample(1:100, 20)
data.uji <- data[ambil.uji, ]
data.model <- data[-ambil.uji, ]
regresi <- lm(Ybbt_badan ~ ., data=data.model)
prediksi.uji <- predict(regresi, data.uji)
MAPE <- mean(abs(data.uji$Ybbt_badan - prediksi.uji)/
data.uji$Ybbt_badan *100)
MAPE
## [1] 1.462459
Program 2.3
data <- read.csv("data sapi.csv")
data <- data[,-1]
regresi <- lm(Ybbt_badan ~ ., data=data)
summary(regresi)
Output 2.2
##
## Call:
## lm(formula = Ybbt_badan ~ ., data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -115.727 -28.822 -0.299 32.171 108.928
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 118.8305 91.9128 1.293 0.199
## X1pjg_badan 6.1877 0.1552 39.857 <2e-16 ***
## X2ttg_badan 4.5001 0.2917 15.425 <2e-16 ***
## X3lkr_badan 7.4794 0.3046 24.557 <2e-16 ***
## X4pjg_ekor 1.1279 1.5674 0.720 0.474
## X5pjg_tanduk 0.1448 1.1915 0.122 0.904
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 45.41 on 94 degrees of freedom
## Multiple R-squared: 0.966, Adjusted R-squared: 0.9642
## F-statistic: 533.5 on 5 and 94 DF, p-value: < 2.2e-16
Program 2.4
rumah <- read.csv("rumah.csv")
modelrumah <- lm(harga ~ luasbangunan + luastanah +
umur + dekattol, data=rumah)
summary(modelrumah)
library(car)
vif(modelrumah)
Output 2.3
##
## Call:
## lm(formula = harga ~ luasbangunan + luastanah + umur + dekattol,
## data = rumah)
##
## Residuals:
## Min 1Q Median 3Q Max
## -104.309 -22.753 -3.994 22.168 119.011
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 195.20528 21.03807 9.279 5.43e-13 ***
## luasbangunan -0.09913 0.43881 -0.226 0.82208
## luastanah 1.28826 0.43634 2.952 0.00457 **
## umur -3.82505 0.51445 -7.435 5.99e-10 ***
## dekattol 22.24126 11.49569 1.935 0.05799 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 38.45 on 57 degrees of freedom
## Multiple R-squared: 0.8534, Adjusted R-squared: 0.8431
## F-statistic: 82.96 on 4 and 57 DF, p-value: < 2.2e-16
## luasbangunan luastanah umur dekattol
## 41.507084 41.509087 1.025362 1.061228
Output 2.4
##
## Call:
## lm(formula = harga ~ luasbangunan + umur + dekattol, data = rumah)
##
## Residuals:
## Min 1Q Median 3Q Max
## -86.602 -26.876 -2.347 22.319 125.661
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 227.02384 19.23260 11.804 < 2e-16 ***
## luasbangunan 1.17990 0.07438 15.863 < 2e-16 ***
## umur -3.74643 0.54687 -6.851 5.22e-09 ***
## dekattol 22.38818 12.23647 1.830 0.0724 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 40.93 on 58 degrees of freedom
## Multiple R-squared: 0.831, Adjusted R-squared: 0.8223
## F-statistic: 95.06 on 3 and 58 DF, p-value: < 2.2e-16
## luasbangunan umur dekattol
## 1.052514 1.022615 1.061208
2.2 Pohon Regresi dan Ensemble-nya
Program 2.5
rumah <- read.csv("rumah.csv")
library(rpart)
pohon <- rpart(harga ~ luasbangunan +
umur + dekattol, data=rumah,
control = rpart.control(minsplit=20, c=0))
library(rpart.plot)
rpart.plot(pohon, extra=1)
Gambar 2.1
Program 2.6
prediksiharga <- predict(pohon, rumah)
MAPE <- mean(abs(rumah$harga - prediksiharga)/
rumah$harga *100)
MAPE
## [1] 13.83895
Gambar 2.2
Program 2.7
library(randomForest)
modelRF <- randomForest(harga ~ luasbangunan +
umur + dekattol, data=rumah,
ntree=500, mtry=3)
prediksiRF <- predict(modelRF, rumah)
MAPE <- mean(abs(rumah$harga - prediksiRF)/
rumah$harga *100)
MAPE
## [1] 6.359173
Program 2.8
library(gbm)
grad.boost <- gbm(
formula = harga ~ luasbangunan +
umur + dekattol,
data = rumah,
n.trees = 300,
shrinkage = 0.1,
interaction.depth = 3,
)
## Distribution not specified, assuming gaussian ...
prediksi.gb <- predict(grad.boost, rumah)
## Using 300 trees...
MAPE <- mean(abs(rumah$harga - prediksi.gb)/
rumah$harga *100)
MAPE
## [1] 9.828887
2.3 Support Vector Regression (SVR)
Program 2.9
library(ISLR)
Auto <- Auto[which(Auto$horsepower > 60),]
plot(Auto$horsepower, Auto$mpg, pch=19, col="grey90", cex=1.5,
xlab="horsepower", ylab="miles per gallon")
points(Auto$horsepower, Auto$mpg, col="grey", cex=1.5)
library(e1071)
#SVR linear
modelsvm = svm(mpg ~ horsepower, Auto, kernel="linear")
predYsvm = predict(modelsvm, Auto)
points(Auto$horsepower, predYsvm, col = "red", pch=19, cex=0.5)
#SVR radial
modelsvm = svm(mpg ~ horsepower, Auto, kernel="radial", cost=2)
predYsvm = predict(modelsvm, Auto)
points(Auto$horsepower, predYsvm, col = "blue", pch=19, cex=0.5)
Program 2.9
2.4 Neural Network untuk Pemodelan Regresi
Program 2.10
library(ISLR)
Auto <- Auto[which(Auto$horsepower > 60),]
data <- Auto[,c("mpg", "horsepower")]
maxs <- apply(data, 2, max)
mins <- apply(data, 2, min)
scaled <- as.data.frame(scale(data, center = mins,
scale = maxs - mins))
library(neuralnet)
nn <- neuralnet(mpg ~ horsepower,
data = scaled, hidden = c(3),
act.fct = "logistic",
linear.output = TRUE)
nn$result.matrix
Output 2.5
## [,1]
## error 2.2007578
## reached.threshold 0.0074052
## steps 164.0000000
## Intercept.to.1layhid1 0.1864301
## horsepower.to.1layhid1 -0.2859610
## Intercept.to.1layhid2 0.4618093
## horsepower.to.1layhid2 -3.8322613
## Intercept.to.1layhid3 -1.0018501
## horsepower.to.1layhid3 -7.0302166
## Intercept.to.mpg -0.3951037
## 1layhid1.to.mpg 0.9829309
## 1layhid2.to.mpg 0.2932757
## 1layhid3.to.mpg 1.4532635
Program 2.11
plot(Auto$horsepower, Auto$mpg, pch=19, col="grey90", cex=1.5,
xlab="horsepower", ylab="miles per gallon")
points(Auto$horsepower, Auto$mpg, col="grey", cex=1.5)
pr.nn <- compute(nn, scaled)$net.result
prediksi.nn <- pr.nn * (max(Auto$mpg)-min(Auto$mpg)) + min(Auto$mpg)
points(Auto$horsepower, prediksi.nn, col="blue", pch=19, cex=0.5)
Gambar 2.7
