Decomposition
There is an increasing trend based on a 12 month moving average and there is seasonal fluctuations in which sales are higher in the summer and lower in the winter.
plastics %>% decompose(type = "multiplicative") -> classical
autoplot(classical) +
labs(title = "Plastic Monthly Sales",
x = "Year", y = "")## Jan Feb Mar Apr May Jun Jul Aug
## 1 NA NA NA NA NA NA 976.9583 977.0417
## Sep Oct Nov Dec
## 1 977.0833 978.4167 982.7083 990.4167
## Jan Feb Mar Apr May Jun Jul
## 1 0.7670466 0.7103357 0.7765294 0.9103112 1.0447386 1.1570026 1.1636317
## Aug Sep Oct Nov Dec
## 1 1.2252952 1.2313635 1.1887444 0.9919176 0.8330834
Yes, the results support the graphical interpretation from part a. The trend is increasing and there is seasonality.
autoplot(plastics, series="Data") +
autolayer(trendcycle(classical), series="Trend") +
autolayer(seasadj(classical), series="Seas. Adj.") +
labs(title = "Plastic Monthly Sales",
x = "Year", y = "") +
scale_color_brewer(palette = "Set2")plastics_with_outlier <- plastics
plastics_with_outlier[30] <- plastics_with_outlier[30] + 500
decompose_plastics_with_outlier <- decompose(plastics_with_outlier, type="multiplicative")
seasonally_adjusted_plastics_with_outlier <- plastics_with_outlier / decompose_plastics_with_outlier$seasonal
autoplot(plastics, series = 'Data') +
autolayer(seasadj(classical), series = 'without outlier') +
autolayer(seasonally_adjusted_plastics_with_outlier, series = 'with outlier') +
labs(title = "Plastic Monthly Sales",
x = "Year", y = "") +
scale_color_brewer(palette = "Set2")There is a difference if the outlier is near the end rather than in the middle of the time series. If the outlier is in the middle of the time series the seasonally-adjusted time series is affected. If the outlier is near either of the ends of the time series the seasonally-adjusted time series would be the only thing affected.
Decompose the series using X11. Does it reveal any outliers, or unusual features that you had not noticed previously?
There are some spikes in the remainder early on (circa 1983) and around 2000. That indicates the presence of some outliers. Yes, the decomposition using X11 reveals Outliers not noticed previously. They are evident in the remainder component specifically, in the early years of the time series, and in the later years.
retaildata <- readxl::read_excel("retail.xlsx", skip=1)
retail <- ts(retaildata[, "A3349337W"], frequency = 12, start = c(1982, 4))
x11_retail <- seas(retail, x11="")
autoplot(x11_retail) +
labs(title="X11 Decomposition")