LOAD PACKAGE

library(forecast)

## Warning: package 'forecast' was built under R version 4.5.3

library(tseries)

## Warning: package 'tseries' was built under R version 4.5.3

## Registered S3 method overwritten by 'quantmod':
##   method            from
##   as.zoo.data.frame zoo

library(readxl)

## Warning: package 'readxl' was built under R version 4.5.3

IMPORT DATA EXCEL

data <- read_excel(
  "~/PROJECT SEMESTER 6/PROJECT STATISTIKA 1/DATA PROJECT UAS JST.xlsx"
)

head(data)

## # A tibble: 6 × 2
##   Tanggal    Terakhir
##   <chr>      <chr>   
## 1 30/12/2025 8.646,94
## 2 29/12/2025 8.644,26
## 3 24/12/2025 8.537,91
## 4 23/12/2025 8.584,78
## 5 22/12/2025 8.645,84
## 6 19/12/2025 8.609,55

MELIHAT STRUKTUR DATA

str(data)

## tibble [474 × 2] (S3: tbl_df/tbl/data.frame)
##  $ Tanggal : chr [1:474] "30/12/2025" "29/12/2025" "24/12/2025" "23/12/2025" ...
##  $ Terakhir: chr [1:474] "8.646,94" "8.644,26" "8.537,91" "8.584,78" ...

summary(data)

##    Tanggal            Terakhir        
##  Length:474         Length:474        
##  Class :character   Class :character  
##  Mode  :character   Mode  :character

DATA CLEANING

# Mengubah format angka Indonesia menjadi numerik

data$Terakhir <- as.numeric(
  gsub(",", ".", 
       gsub("\\.", "", data$Terakhir))
)

# Mengubah tanggal menjadi format Date

data$Tanggal <- as.Date(
  data$Tanggal,
  format = "%d/%m/%Y"
)

# Mengurutkan data berdasarkan tanggal

data <- data[order(data$Tanggal), ]

# Mengecek missing value

sum(is.na(data$Terakhir))

## [1] 0

head(data)

## # A tibble: 6 × 2
##   Tanggal    Terakhir
##   <date>        <dbl>
## 1 2024-01-15    7224 
## 2 2024-01-16    7243.
## 3 2024-01-17    7201.
## 4 2024-01-18    7253.
## 5 2024-01-19    7227.
## 6 2024-01-22    7248.

MEMBUAT TIME SERIES

ts_data <- ts(data$Terakhir)

plot(
  ts_data,
  main = "Plot Data IHSG",
  ylab = "Harga Penutupan",
  xlab = "Waktu",
  col = "blue"
)

# UJI STASIONERITAS AWAL

adf_awal <- adf.test(ts_data)

adf_awal

## 
##  Augmented Dickey-Fuller Test
## 
## data:  ts_data
## Dickey-Fuller = -1.9426, Lag order = 7, p-value = 0.6023
## alternative hypothesis: stationary

Interpretasi: Jika p-value > 0.05 maka data tidak stasioner

DIFFERENCING

diff1 <- diff(ts_data)

plot(
  diff1,
  main = "Differencing Orde 1",
  ylab = "Differencing",
  xlab = "Waktu",
  col = "red"
)

adf_diff <- adf.test(diff1)

## Warning in adf.test(diff1): p-value smaller than printed p-value

adf_diff

## 
##  Augmented Dickey-Fuller Test
## 
## data:  diff1
## Dickey-Fuller = -7.2105, Lag order = 7, p-value = 0.01
## alternative hypothesis: stationary

PLOT ACF DAN PACF

acf(
  diff1,
  main = "Plot ACF"
)

pacf(
  diff1,
  main = "Plot PACF"
)

# MEMBAGI DATA TRAINING DAN TESTING

n <- length(ts_data)

train_size <- floor(0.8 * n)

train <- ts_data[1:train_size]

test <- ts_data[(train_size + 1):n]

length(train)

## [1] 379

length(test)

## [1] 95

PEMODELAN ARIMA

model_arima <- auto.arima(
  train,
  seasonal = FALSE
)

summary(model_arima)

## Series: train 
## ARIMA(0,1,0) 
## 
## sigma^2 = 10429:  log likelihood = -2285.06
## AIC=4572.11   AICc=4572.12   BIC=4576.05
## 
## Training set error measures:
##                      ME     RMSE      MAE         MPE      MAPE      MASE
## Training set -0.3596464 101.9897 67.13956 -0.01570036 0.9305611 0.9976447
##                    ACF1
## Training set -0.0276924

FORECASTING ARIMA

forecast_arima <- forecast(
  model_arima,
  h = length(test)
)

plot(
  forecast_arima,
  main = "Forecasting ARIMA"
)

# EVALUASI MODEL ARIMA

akurasi_arima <- accuracy(
  forecast_arima,
  test
)

akurasi_arima

##                       ME     RMSE       MAE         MPE      MAPE      MASE
## Training set  -0.3596464 101.9897  67.13956 -0.01570036 0.9305611 0.9976447
## Test set     225.7737895 303.3168 249.63779  3.01616523 3.3615990 3.7094347
##                    ACF1
## Training set -0.0276924
## Test set             NA

PEMODELAN SARIMA

model_sarima <- auto.arima(
  train,
  seasonal = TRUE
)

summary(model_sarima)

## Series: train 
## ARIMA(0,1,0) 
## 
## sigma^2 = 10429:  log likelihood = -2285.06
## AIC=4572.11   AICc=4572.12   BIC=4576.05
## 
## Training set error measures:
##                      ME     RMSE      MAE         MPE      MAPE      MASE
## Training set -0.3596464 101.9897 67.13956 -0.01570036 0.9305611 0.9976447
##                    ACF1
## Training set -0.0276924

FORECASTING SARIMA

forecast_sarima <- forecast(
  model_sarima,
  h = length(test)
)

plot(
  forecast_sarima,
  main = "Forecasting SARIMA"
)

# EVALUASI MODEL SARIMA

akurasi_sarima <- accuracy(
  forecast_sarima,
  test
)

akurasi_sarima

##                       ME     RMSE       MAE         MPE      MAPE      MASE
## Training set  -0.3596464 101.9897  67.13956 -0.01570036 0.9305611 0.9976447
## Test set     225.7737895 303.3168 249.63779  3.01616523 3.3615990 3.7094347
##                    ACF1
## Training set -0.0276924
## Test set             NA

TABEL PERBANDINGAN MODEL

perbandingan <- data.frame(

  Model = c(
    "ARIMA",
    "SARIMA"
  ),

  RMSE = c(
    akurasi_arima[2, "RMSE"],
    akurasi_sarima[2, "RMSE"]
  ),

  MAE = c(
    akurasi_arima[2, "MAE"],
    akurasi_sarima[2, "MAE"]
  ),

  MAPE = c(
    akurasi_arima[2, "MAPE"],
    akurasi_sarima[2, "MAPE"]
  ),

  AIC = c(
    AIC(model_arima),
    AIC(model_sarima)
  )
)

perbandingan

##    Model     RMSE      MAE     MAPE      AIC
## 1  ARIMA 303.3168 249.6378 3.361599 4572.114
## 2 SARIMA 303.3168 249.6378 3.361599 4572.114

MENENTUKAN MODEL TERBAIK

if(
  min(perbandingan$RMSE) ==
  perbandingan$RMSE[1]
){

  cat("Model terbaik adalah ARIMA")

} else {

  cat("Model terbaik adalah SARIMA")

}

## Model terbaik adalah ARIMA

DIAGNOSTIK RESIDUAL

checkresiduals(model_arima)

## 
##  Ljung-Box test
## 
## data:  Residuals from ARIMA(0,1,0)
## Q* = 8.9193, df = 10, p-value = 0.5398
## 
## Model df: 0.   Total lags used: 10

checkresiduals(model_sarima)

## 
##  Ljung-Box test
## 
## data:  Residuals from ARIMA(0,1,0)
## Q* = 8.9193, df = 10, p-value = 0.5398
## 
## Model df: 0.   Total lags used: 10

FORECAST 30 HARI KE DEPAN

forecast_30 <- forecast(
  model_arima,
  h = 30
)

plot(
  forecast_30,
  main = "Forecast 30 Hari Kedepan"
)

forecast_30

##     Point Forecast    Lo 80    Hi 80    Lo 95    Hi 95
## 380        7080.47 6949.592 7211.348 6880.310 7280.630
## 381        7080.47 6895.381 7265.559 6797.400 7363.540
## 382        7080.47 6853.783 7307.157 6733.782 7427.158
## 383        7080.47 6818.714 7342.226 6680.149 7480.791
## 384        7080.47 6787.818 7373.122 6632.898 7528.042
## 385        7080.47 6759.886 7401.054 6590.179 7570.761
## 386        7080.47 6734.200 7426.740 6550.895 7610.045
## 387        7080.47 6710.292 7450.648 6514.331 7646.609
## 388        7080.47 6687.836 7473.104 6479.989 7680.951
## 389        7080.47 6666.598 7494.342 6447.507 7713.433
## 390        7080.47 6646.397 7514.543 6416.613 7744.327
## 391        7080.47 6627.096 7533.844 6387.094 7773.846
## 392        7080.47 6608.583 7552.357 6358.782 7802.158
## 393        7080.47 6590.770 7570.170 6331.538 7829.402
## 394        7080.47 6573.582 7587.358 6305.252 7855.688
## 395        7080.47 6556.959 7603.981 6279.829 7881.111
## 396        7080.47 6540.847 7620.093 6255.188 7905.752
## 397        7080.47 6525.202 7635.738 6231.261 7929.679
## 398        7080.47 6509.987 7650.953 6207.991 7952.949
## 399        7080.47 6495.167 7665.773 6185.326 7975.614
## 400        7080.47 6480.712 7680.228 6163.220 7997.720
## 401        7080.47 6466.599 7694.341 6141.635 8019.305
## 402        7080.47 6452.802 7708.138 6120.535 8040.405
## 403        7080.47 6439.302 7721.638 6099.888 8061.052
## 404        7080.47 6426.081 7734.859 6079.668 8081.272
## 405        7080.47 6413.121 7747.819 6059.848 8101.092
## 406        7080.47 6400.409 7760.531 6040.406 8120.534
## 407        7080.47 6387.930 7773.010 6021.321 8139.619
## 408        7080.47 6375.671 7785.269 6002.573 8158.367
## 409        7080.47 6363.623 7797.317 5984.146 8176.794

Perbandingan Model ARIMA dan SARIMA pada Forecasting Indeks Harga Saham Gabungan (IHSG)

Gracelinda Chrisageulis

2026-05-29

LOAD PACKAGE

IMPORT DATA EXCEL

MELIHAT STRUKTUR DATA

DATA CLEANING

MEMBUAT TIME SERIES

DIFFERENCING

PLOT ACF DAN PACF

PEMODELAN ARIMA

FORECASTING ARIMA

PEMODELAN SARIMA

FORECASTING SARIMA

TABEL PERBANDINGAN MODEL

MENENTUKAN MODEL TERBAIK

DIAGNOSTIK RESIDUAL

FORECAST 30 HARI KE DEPAN