## ----include = FALSE----------------------------------------------------------
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.width = 7,
  fig.height = 4.5,
  fig.align = "center",
  message = FALSE,
  warning = FALSE
)

## ----setup--------------------------------------------------------------------
library(trendseries)
library(dplyr)
library(tidyr)
library(ggplot2)

## ----libs---------------------------------------------------------------------
# library(trendseries)
# library(dplyr)
# library(tidyr)

## ----theme--------------------------------------------------------------------
library(ggplot2)

theme_series <- theme_minimal(paper = "#fefefe") +
  theme(
    legend.position = "bottom",
    panel.grid.minor = element_blank(),
    strip.background = element_rect(fill = "#2c3e50"),
    strip.text = element_text(color = "#fefefe"),
    axis.ticks.x = element_line(color = "gray40", linewidth = 0.5),
    axis.line.x = element_line(color = "gray40", linewidth = 0.5),
    axis.title.x = element_blank(),
    palette.colour.discrete = c(
      "#2c3e50",
      "#e74c3c",
      "#f39c12",
      "#1abc9c",
      "#9b59b6"
    )
  )

## ----gdp-plot-----------------------------------------------------------------
ggplot(gdp_construction, aes(date, index)) +
  geom_line(lwd = 0.7) +
  scale_x_date(date_breaks = "2 years", date_labels = "%Y") +
  labs(
    title = "Brazilian construction activity",
    y = "Index"
  ) +
  theme_series

## ----gdp-decomp---------------------------------------------------------------
gdp_parts <- gdp_construction |>
  decompose_series(value_col = "index")

gdp_parts

## ----gdp-long-----------------------------------------------------------------
gdp_long <- gdp_parts |>
  pivot_longer(
    cols = c(index, trend_stl, seasonal_stl, remainder_stl),
    names_to = "component",
    values_to = "value"
  ) |>
  mutate(
    component = factor(
      component,
      levels = c("index", "trend_stl", "seasonal_stl", "remainder_stl"),
      labels = c("Observed", "Trend", "Seasonal", "Remainder")
    )
  )

## ----gdp-facet----------------------------------------------------------------
ggplot(gdp_long, aes(date, value)) +
  geom_line(aes(color = component), lwd = 0.7, show.legend = FALSE) +
  facet_wrap(vars(component), ncol = 1, scales = "free_y") +
  scale_x_date(date_breaks = "2 years", date_labels = "%Y") +
  labs(
    title = "STL decomposition of construction activity",
    subtitle = "Observed = Trend + Seasonal + Remainder",
    y = NULL
  ) +
  theme_series

## ----stl-params, eval = FALSE-------------------------------------------------
# # Allow the seasonal pattern to evolve slowly over time, and downweight outliers
# gdp_robust <- gdp_construction |>
#   decompose_series(
#     value_col = "index",
#     params = list(s.window = 13, robust = TRUE)
#   )

## ----reg-decomp---------------------------------------------------------------
suboil <- oil_derivatives |>
  filter(between(date, as.Date("1995-06-01"), as.Date("2006-12-01"))) |>
  mutate(lprod = log(production))

ggplot(suboil, aes(date, lprod)) +
  geom_line(lwd = 0.7) +
  labs(
    title = "Petroleum derivatives production",
    y = "Thousand barrels per day (log scale)"
  ) +
  theme_series

## -----------------------------------------------------------------------------
suboil_reg <- suboil |>
  decompose_series(
    value_col = "lprod",
    methods = "regression",
    trend = "cubic"
  )

ggplot(suboil_reg, aes(date)) +
  geom_line(aes(y = lprod, color = "Original"), lwd = 0.7) +
  geom_line(aes(y = trend_regression, color = "Trend (reg.)"), lwd = 0.7) +
  labs(title = "Regression cubic trend", y = "log production", color = NULL) +
  theme_series

## -----------------------------------------------------------------------------
ggplot(suboil_reg, aes(date, remainder_regression)) +
  geom_line(lwd = 0.7) +
  labs(title = "Regression remainder", y = NULL) +
  theme_series

## ----compare-trends-----------------------------------------------------------
comparison <- suboil |>
  decompose_series(value_col = "lprod", methods = "stl") |>
  decompose_series(value_col = "lprod", methods = "regression", trend = "cubic")

comparison_long <- comparison |>
  pivot_longer(
    cols = trend_stl:remainder_regression,
    names_pattern = "(.*)_(.*)",
    names_to = c("component", "method"),
    values_to = "trend"
  ) |>
  mutate(
    component = factor(component, levels = c("trend", "seasonal", "remainder"))
  )

ggplot(comparison_long, aes(date, trend)) +
  # Original series
  geom_line(
    data = suboil,
    aes(y = lprod),
    layout = c(1, 2),
    lwd = 0.5,
    alpha = 0.5
  ) +
  # Components and methods
  geom_line(aes(y = trend, color = component), lwd = 0.7) +
  facet_grid(vars(component), vars(method), scales = "free_y") +
  guides(color = "none") +
  labs(title = "STL vs regression decomposition", y = NULL) +
  theme_series

## ----elec-decomp--------------------------------------------------------------
electricity_parts <- electricity |>
  mutate(lvalue = log(value)) |>
  decompose_series(value_col = "lvalue", group_cols = "name_series")

glimpse(electricity_parts)

## ----elec-plot----------------------------------------------------------------
ggplot(electricity_parts, aes(date)) +
  geom_line(aes(y = seasonal_stl), color = "#2c3e50", lwd = 0.8) +
  facet_wrap(vars(name_series), ncol = 1, scales = "free_y") +
  scale_x_date(date_breaks = "3 years", date_labels = "%Y") +
  labs(
    title = "Electricity consumption by sector",
    subtitle = "STL seasonal component extracted per group (log scale)",
    y = "seasonal factor (log)"
  ) +
  theme_series

## -----------------------------------------------------------------------------
# decompose_series(
#   oil_derivatives,
#   value_col = "production",
#   transform = "log"
# )

## -----------------------------------------------------------------------------
# decompose_series(
#   oil_derivatives,
#   value_col = "production",
#   methods = "classic",
#   transform = "log"
# )

## -----------------------------------------------------------------------------
# decompose_series(
#   suboil,
#   value_col = "lprod",
#   methods = "bsm"
# )

## ----seats, eval = requireNamespace("seasonal", quietly = TRUE)---------------
# requires the 'seasonal' package
seas_oil <- decompose_series(
  suboil,
  value_col = "lprod",
  methods = "seats"
)

ggplot(seas_oil, aes(date)) +
  geom_line(aes(y = lprod, color = "Original"), lwd = 0.7) +
  geom_line(aes(y = trend_seats, color = "Trend (SEATS)"), lwd = 0.7) +
  labs(title = "SEATS trend-cycle", y = "log production", color = NULL) +
  theme_series

## ----seats-seasonal, eval = requireNamespace("seasonal", quietly = TRUE)------
ggplot(seas_oil, aes(date, seasonal_seats)) +
  geom_line(lwd = 0.7) +
  labs(title = "SEATS seasonal component", y = NULL) +
  theme_series

## ----deseason, eval = requireNamespace("seasonal", quietly = TRUE)------------
deseas_oil <- deseason_series(
  suboil,
  value_col = "lprod",
  methods = "seats",
  # set to TRUE to also return the trend, seasonal, and remainder components
  components = FALSE
)

ggplot(deseas_oil, aes(date, seasadj_seats)) +
  geom_line(lwd = 0.7) +
  labs(
    title = "SEATS seasonally adjusted series",
    y = "log production"
  ) +
  theme_series

## -----------------------------------------------------------------------------
# decompose_series(
#   suboil,
#   value_col = "lprod",
#   methods = "seats",
#   seasadj = TRUE
# )