data-raw/aemo.R

In robjhyndman/fpp: Data for "Forecasting: Principles and Practice" (2nd Edition)

# AEMO

library(dplyr)
library(lubridate)
library(forecast)

aemo <- readr::read_csv("data-raw/aemo2015/demand_VIC.csv")
temp <- readr::read_csv("data-raw/aemo2015/temp_86071.csv")
aemo$Date <- as.Date(aemo$Date, origin = "1899-12-30")
temp$Date <- as.Date(temp$Date, origin = "1899-12-30")
firstday <- as.Date("2014-01-01", origin = "1899-12-30")
lastday <- as.Date("2014-12-31", origin = "1899-12-30")

aemo <- filter(aemo, Date >= firstday & Date <= lastday)
temp <- filter(temp, Date >= firstday & Date <= lastday)

hols <- as.Date(c(
  "2014-01-01",
  "2014-01-27",
  "2014-03-10",
  "2014-04-18",
  "2014-04-21",
  "2014-04-25",
  "2014-06-09",
  "2014-11-04",
  "2014-12-25",
  "2014-12-26"
))
pubhol <- is.element(aemo$Date, hols)
dow <- wday(aemo$Date, label = TRUE)
workday <- dow != "Sun" & dow != "Sat" & !pubhol
demand <- cbind(
  Demand = aemo$OperationalLessIndustrial / 1e3,
  WorkDay = workday,
  Temperature = temp$Temp
)
demand <- msts(
  demand,
  seasonal.periods = c(48, 48 * 7, 48 * 365),
  start = 2014
)
plot(demand)
table(demand[, 2]) / 48
elecdemand <- demand
usethis::use_data(elecdemand, overwrite = TRUE)

# elecdaily
elecdaily <- cbind(
  aggregate.ts(elecdemand[, "Demand"], nfrequency = 365),
  aggregate.ts(elecdemand[, c("WorkDay")], nfrequency = 365, mean),
  aggregate.ts(elecdemand[, c("Temperature")], nfrequency = 365, max)
)
colnames(elecdaily) <- c("Demand", "WorkDay", "Temperature")
elecdaily <- ts(elecdaily, start = c(1,4), frequency = 7)
usethis::use_data(elecdaily, overwrite = TRUE)