aemo_predispatch: Price and demand forecasts (P5MIN, PREDISPATCH)

Description

Returns AEMO's forecast prices and demand for a NEM region. Two horizons:

"p5min": 5-minute-ahead forecast, 12 intervals ahead, published every 5 minutes.
"predispatch": 40-hour-ahead predispatch at 30-minute resolution, published every 30 minutes.

Usage

aemo_predispatch(
  region,
  start,
  end,
  horizon = c("predispatch", "p5min"),
  run_datetime = NULL
)

Value

An aemo_tbl.

Arguments

region: NEM region code.
start, end: Window of forecast run-times.
horizon: One of "predispatch" (default) or "p5min".
run_datetime: Optional character or POSIXct. A specific RUN_DATETIME to pin to. NULL (default) returns every vintage issued in [start, end].

Details

The 7-day predispatch publication was retired when 5-minute settlement commenced; for longer horizons use aemo_pasa().

Vintages. PREDISPATCH and P5MIN forecasts are re-issued every 30 or 5 minutes respectively. Every vintage is archived by its RUN_DATETIME. By default (run_datetime = NULL) this function returns all vintages whose file timestamp falls in [start, end], i.e. all the forecasts issued during the window. The periodid / datetime columns on the returned rows give each forecast's target time.

For forecast-error research (comparing a forecast vintage against the realised dispatch) pass run_datetime to pin a specific vintage:

# The PREDISPATCH run issued at 15:00 on 1 June 2024 --
# 80 half-hour-ahead rows covering 15:30 out to 31:30 hours.
v <- aemo_predispatch("NSW1", start = "2024-06-01", end = "2024-06-02",
                      run_datetime = "2024-06-01 15:00")

This is the vintage-aware pattern used in Prakash (2023) NEMSEER (JOSS 8(92) 5883).

Examples

Run this code

# \donttest{
op <- options(aemo.cache_dir = tempdir())
try({
  now <- Sys.time()
  p <- aemo_predispatch("NSW1", start = now - 3600, end = now)
  head(p)
})
options(op)
# }