Nadja Klein, Michael Stanley Smith, and David J. Nott, "Deep
Distributional Time Series Models and the Probabilistic Forecasting of
Intraday Electricity Prices", Journal of Applied Econometrics,
Vol. 38, No. 4, 2023, pp. 493-511.

The file "electricity.csv" contains all the electricity data used to
fit the time-series models for the 5 regional price series in Section
5, plus the NSW price time series models which use demand forecasts in
Section 6. There are a total of 19,152 observations.

This file is an ASCII file in DOS format. It is zipped in the file 
ksn-data.zip. Unix/Linux users should use "unzip -a".

The variables are:

1. INTERVAL_DATETIME: the date and time at the beginning of the half-hour
period.

2. OPERATIONAL_DEMAND_POE10: the 24-hour ahead forecast of the 10th
percentile of NSW total system demand. In the NEM "POE" is shorthand for
"probability of exceedance".

3. OPERATIONAL_DEMAND_POE50: the 24-hour ahead forecast of the 50th
percentile (i.e. median) of NSW total system demand.

4. OPERATIONAL_DEMAND_POE90: the 24-hour ahead forecast of the 90th
percentile of NSW total system demand.

5. Diff_dt: the number of hours ahead the forecast made, computed as
the difference between LASTCHANGED (see additional notes below) and
the start of the half-hourly period. Computed as a final double-check
on the data and it is a redundant variable unused in the paper.

6. NSW1, QLD1, SA1, TAS1, VIC1: electricity spot prices (Australian
dollars per MWh) during the half-hourly trading intervals in each of
the five regions. 


Additional Notes Concerning Source and Construction of the Data

A. Electricity Prices

The electricity price data are extracted from the files called
"PRICE_AND_DEMAND_2019XXXX_SSSS.csv" that were sourced from AMEO's
publicly available data at the time of writing, where SSS is the
region (i.e. NSW1, QLD1, VIC1, TAS1, SA1) and XXXX is the month and
day.

B. 24 Hour Ahead Electricity Demand Forecasts

The demand forecast data are extracted from a large number of files
entitled "PUBIC_FORECAST_DEMAND_HH_2019XX_2019XX.csv" which were made
available at the website
"nemweb.com.au/#operational-demand-forecast-hh" at the time of
writing. 

In these files, the three demand forecast variables are labeled
OPERATIONAL_DEMAND_POE10, OPERATIONAL_DEMAND_POE50 and
OPERATIONAL_DEMAND_POE90.

There is one file per half-hourly forecast origin (labeled using the
variable LASTCHANGED), so 48 files are generated each day. From each
of the 365 x 48 files generated in 2019, we extract the single
forecast that is exactly 24 hours ahead for region NSW only. We then
match it with the price data above. These define the three NSW
operational demand forecast variables above.