# Input data
Data     <- read_excel("total_kunjungan_mancanegara.xlsx")
Visitors <- Data$Visitors
Visitors.Diff1 <- diff(Visitors, differences = 1)
# Plot hasil differencing
ts.plot(Visitors.Diff1, col= "darkgreen", main="Time Series Plot")

# Transformasi
Visitors.Log.Diff1 <- diff(log(Visitors), diff=1)
ts.plot(Visitors.Log.Diff1, col="darkgreen", main="Time Series Plot")

Terlihat transformasi log dari data sebelum dilakukan difference, variansi data relatif stabil namun terdapat 1 outlier.

# Identifikasi
par(mfrow=c(1,2))
acf(Visitors.Log.Diff1, 
    lag.max = 33, 
    type = c("correlation","covariance","partial"), 
    plot = TRUE, 
    na.action = na.pass)

pacf(Visitors.Log.Diff1, 
     lag.max = 33, 
     na.action = na.pass)

Dapat dilihat diatas kalau fungsi ACF signifikan pada Lag 1 dan 2, dan PACF signifikan pada lag 1.

Visitors.log <- log(Visitors)
arimamodel1  <- arima(Visitors.log, 
                      order = c(1,1,1), 
                      seasonal = list(order = c(0,0,0), period = NA), 
                      include.mean = FALSE)

summary(arimamodel1)
## 
## Call:
## arima(x = Visitors.log, order = c(1, 1, 1), seasonal = list(order = c(0, 0, 
##     0), period = NA), include.mean = FALSE)
## 
## Coefficients:
##          ar1      ma1
##       0.4826  -0.0578
## s.e.  0.2918   0.3212
## 
## sigma^2 estimated as 0.0475:  log likelihood = 3.24,  aic = -0.48
## 
## Training set error measures:
##                       ME      RMSE       MAE        MPE      MAPE     MASE
## Training set -0.03342497 0.2146271 0.1312295 -0.2620017 0.9962399 1.029327
##                     ACF1
## Training set -0.01460957