data-raw/US_change/uschange.R

In robjhyndman/fpp3-package: Data for "Forecasting: Principles and Practice" (3rd Edition)

library(tidyverse)
library(tsibble)

# US change data
# Downloaded from https://research.stlouisfed.org/pdl/1216/

# Read monthly unemployment data
unrate <- read_tsv(here::here("data-raw/US_change/uschange_fpp3_Monthly.txt")) |>
  mutate(Month = yearmonth(observation_date)) |>
  select(Month, everything(), -observation_date) |>
  rename(Unemployment = UNRATE_20191004) |>
  as_tsibble(index = Month)

# Select last month of each quarter
unrate <- unrate |>
  mutate(Quarter = yearquarter(Month)) |>
  index_by(Quarter) |>
  summarise(Unemployment = tail(Unemployment, 1)) |>
  filter_index("1969 Q4" ~ .)

# Read remaining quarterly data
other <- read_tsv(here::here("data-raw/US_change/uschange_fpp3_Quarterly.txt"), na=c("",".")) |>
  mutate(Quarter = yearquarter(observation_date)) |>
  select(Quarter, everything(), -observation_date) |>
  rename(
    Income = DPIC96_20191004,
    Production = IPB50001SQ_20191004,
    Consumption = PCECC96_20191004,
    Savings = PSAVE_20191004
  ) |>
  as_tsibble(index = Quarter) |>
  filter_index("1969 Q4" ~ .)

# Joint the two tsibbles and compute percentage changes
# need to take differences of the logs * 100 to have growth rates of these.
us_change <- inner_join(other, unrate) |>
  mutate(
    Consumption = difference(log(Consumption)) * 100,
    Income = difference(log(Income)) * 100,
    Production = difference(log(Production)) * 100,
    Savings = difference(log(Savings)) * 100,
    Unemployment = difference(Unemployment)
  ) |>
  filter_index("1970 Q1" ~ .) |>
  select(Quarter, Consumption, Income, Production, Savings, Unemployment)

usethis::use_data(us_change, overwrite = TRUE)