ANALISIS REGRESI LINEAR SEDERHANA

setwd("E:/KULIAH/SEMESTER 6/(KAMIS) KOMPUTASI STATISTIKA")
dataset   <- read.csv("H051201060_DS_FINAL_KOMSTAT.csv")
dataset

##      x1    x2   x3   y
## 1  7.95  9.23 9.00 3.5
## 2  6.87 10.35 7.35 3.6
## 3  7.65  8.45 7.79 3.2
## 4  7.54  9.61 8.18 4.2
## 5  7.50  8.78 8.84 3.0
## 6  8.15  7.89 8.36 3.0
## 7  7.71  9.22 8.15 3.7
## 8  8.62  8.39 8.40 3.1
## 9  8.37  8.66 8.60 3.6
## 10 8.72  9.10 8.05 3.3
## 11 8.12  6.88 8.78 3.9
## 12 7.40  8.83 9.14 3.3
## 13 7.75  8.91 7.85 3.0
## 14 7.66  9.51 7.87 3.5
## 15 7.88  9.05 8.46 2.8

data_set <- sqrt(dataset)
y         <- data_set$y
x1        <- data_set$x1
x2        <- data_set$x2
x3        <- data_set$x3

UJI LINEARITAS

Ho : Model Linear H1 : Model Tidak Linear

P-Value < 0,05

library(lmtest)
resettest(y~x1)

## 
##  RESET test
## 
## data:  y ~ x1
## RESET = 0.063378, df1 = 2, df2 = 11, p-value = 0.9389

resettest(y~x2)

## 
##  RESET test
## 
## data:  y ~ x2
## RESET = 4.7464, df1 = 2, df2 = 11, p-value = 0.03265

resettest(y~x3)

## 
##  RESET test
## 
## data:  y ~ x3
## RESET = 0.074771, df1 = 2, df2 = 11, p-value = 0.9284

Karena nilai p-value x1, x2, dan x3 > 0,5 maka terima Ho atau tidak cukup bukti untuk menolak Ho. Artinya, ketiga variabel bebas linear terhadap variabel y.

UJI KORELASI

dimana: 0 - 0,2 = sangat lemah 0,3 - 0,4 = lemah 0,5 - 0,6 = kuat 0,7 - 0,8 = sangat kuat

cor(data_set) #fokus ke variabel Y

##            x1         x2          x3           y
## x1  1.0000000 -0.5106162  0.28100211 -0.16989143
## x2 -0.5106162  1.0000000 -0.46391399  0.11371519
## x3  0.2810021 -0.4639140  1.00000000 -0.05563645
## y  -0.1698914  0.1137152 -0.05563645  1.00000000

UJI NORMALITAS

Kriteria penolakan Ho yakni ketika p-value < 0.5

shapiro.test(y)

## 
##  Shapiro-Wilk normality test
## 
## data:  y
## W = 0.9672, p-value = 0.8147

shapiro.test(x1)

## 
##  Shapiro-Wilk normality test
## 
## data:  x1
## W = 0.97086, p-value = 0.8706

shapiro.test(x2)

## 
##  Shapiro-Wilk normality test
## 
## data:  x2
## W = 0.93236, p-value = 0.2959

shapiro.test(x3)

## 
##  Shapiro-Wilk normality test
## 
## data:  x3
## W = 0.98111, p-value = 0.9765

UJI HOMOGENITAS

Kriteria penolakan Ho yakni ketika p-value < 0.5

bartlett.test(data_set)

## 
##  Bartlett test of homogeneity of variances
## 
## data:  data_set
## Bartlett's K-squared = 4.0195, df = 3, p-value = 0.2594

UJI AUTOKORELASI

Kriteria penolakan Ho yakni ketika p-value < 0.5

dwtest(y~x1+x2+x3)

## 
##  Durbin-Watson test
## 
## data:  y ~ x1 + x2 + x3
## DW = 2.4339, p-value = 0.7998
## alternative hypothesis: true autocorrelation is greater than 0

UJI MULTIKOLINERITAS

Yakni ketika nilai VIF > 10

library(car)
vif(lm(y~x1+x2+x3))

##       x1       x2       x3 
## 1.357238 1.592881 1.278526

ANALISIS REGRESI LINEAR

reg   <- lm(y~x1+x3)

x     <- cbind(x0 = array(1, dim = length(y)), x1,x3)
b     <- solve(t(x) %*% x) %*% (t(x)) %*% y
b

##           [,1]
## x0  2.42757970
## x1 -0.20067425
## x3 -0.01022803

summary(reg)

## 
## Call:
## lm(formula = y ~ x1 + x3)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.16119 -0.07410  0.01061  0.06005  0.20210 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)  
## (Intercept)  2.42758    1.20856   2.009   0.0676 .
## x1          -0.20067    0.35516  -0.565   0.5825  
## x3          -0.01023    0.35361  -0.029   0.9774  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.1102 on 12 degrees of freedom
## Multiple R-squared:  0.02893,    Adjusted R-squared:  -0.1329 
## F-statistic: 0.1788 on 2 and 12 DF,  p-value: 0.8385

y_topi  <- reg$coefficients[2] * 8.2
y_topi

##        x1 
## -1.645529