Persiapan Data
data <- read.csv("D:/R/Time Series/Data Revenue & Package 2019-2022.csv", header = TRUE)#data_test
data <- within(data, {
berat <- NULL
revenue <- NULL
})
head(data)
## periode volume
## 1 2019-01-01 6977
## 2 2019-01-02 84942
## 3 2019-01-03 88100
## 4 2019-01-04 78848
## 5 2019-01-05 58016
## 6 2019-01-06 12541
data <- data %>%
mutate(date = ymd(periode))
range(data$date)
## [1] "2019-01-01" "2023-01-12"
## periode volume date
## 0 0 0
Membentuk Data Time Series
Cek Dekomposisi Data
# Membagi Data Training dan Data Testing
data_train <- head(volume_ts, 1426)
data_test <- tail(volume_ts, length(volume_ts)-length(data_train))
#This will plot the time series
ts.plot(data_train, xlab="Periode", ylab="Jumlah Volume", main="Volume Harian Sicepat, 2019-2022")
# This will fit in a line
abline(reg=lm(data_train~time(data_train)))
# Dekomposisi Time Series
data_train%>% decompose() %>% autoplot()
Pemodelan
Holt’s Winters Exponential (Triple Exponential
Smoothing)
# Pemodelan Holt's Winter Exponential
model_hoW <- HoltWinters(data_train)
# Evaluasi Model
model_how_forecast <- forecast(object = model_hoW, h = 43)
volume_ts %>% autoplot(series = "data_train")+
autolayer(data_test, series = "data_test") +
autolayer(model_how_forecast$mean, series = "forecast")
accuracy(model_how_forecast$mean, x = data_test)
## ME RMSE MAE MPE MAPE ACF1 Theil's U
## Test set -107063.3 237543.4 200031.1 -37.38803 46.83678 0.3155771 0.3229196
# Model STLM
model_stlm <- stlm(y = data_train, s.window = 43, method = "ets")
model_stlm_forecast <- forecast(model_stlm, h = 43)
volume_ts %>% autoplot(series = "Data Training")+
autolayer(data_test, series = "Data Testing") +
autolayer(model_stlm_forecast$mean, series = "forecast")
accuracy(model_stlm_forecast$mean, x = data_test)
## ME RMSE MAE MPE MAPE ACF1 Theil's U
## Test set -76642.86 232119.5 191845.3 -34.38522 46.03675 0.2836827 0.3581627
Model Terbaik
Hasil <- data.frame(
Forecast = model_how_forecast$mean,
Volume_Real = data_test
)
Percentage <- with(Hasil, (abs(Forecast - Volume_Real)/Volume_Real)*100)
Fore_per_Act <- with(Hasil, (Forecast/Volume_Real))
Fore_per_Act <- round(Fore_per_Act, 2)
Percentage <- round(Percentage, 2)
#Periode <- data.frame(as.matrix(Percentage), date=time(Percentage))
#Periode <- Periode[,-1]
Hasil_Forecast <- data.frame(Hasil,Fore_per_Act, Percentage)
Hasil_Forecast
## Forecast Volume_Real Fore_per_Act Percentage
## 1 630856.3 756184 0.83 16.57
## 2 682490.4 763572 0.89 10.62
## 3 501733.4 650713 0.77 22.89
## 4 669687.0 290657 2.30 130.40
## 5 1153810.0 1054729 1.09 9.39
## 6 829390.9 870612 0.95 4.73
## 7 840897.3 796187 1.06 5.62
## 8 723865.5 675026 1.07 7.24
## 9 907401.8 672990 1.35 34.83
## 10 791596.9 606972 1.30 30.42
## 11 563543.2 249059 2.26 126.27
## 12 955951.8 1358276 0.70 29.62
## 13 739814.8 1238655 0.60 40.27
## 14 833254.0 880731 0.95 5.39
## 15 665063.6 673929 0.99 1.32
## 16 767227.9 572577 1.34 34.00
## 17 661667.8 482464 1.37 37.14
## 18 405790.3 171566 2.37 136.52
## 19 882077.3 753151 1.17 17.12
## 20 664185.9 570820 1.16 16.36
## 21 734448.2 533562 1.38 37.65
## 22 597132.5 502364 1.19 18.86
## 23 716941.9 497000 1.44 44.25
## 24 627022.6 433156 1.45 44.76
## 25 470144.8 190734 2.46 146.49
## 26 763139.2 930048 0.82 17.95
## 27 737230.8 761210 0.97 3.15
## 28 864351.4 653428 1.32 32.28
## 29 673185.2 596132 1.13 12.93
## 30 757842.8 518991 1.46 46.02
## 31 701531.5 351566 2.00 99.54
## 32 495669.5 143666 3.45 245.02
## 33 898226.9 1021083 0.88 12.03
## 34 526456.0 857401 0.61 38.60
## 35 1124475.6 708520 1.59 58.71
## 36 1001567.2 655426 1.53 52.81
## 37 972889.7 600668 1.62 61.97
## 38 834092.4 505451 1.65 65.02
## 39 592669.3 196722 3.01 201.27
## 40 924922.1 852717 1.08 8.47
## 41 724631.5 709323 1.02 2.16
## 42 846835.5 654272 1.29 29.43
## 43 726995.1 616674 1.18 17.89