nem-data

A simple & opinionated Python command line tool to access Australian National Energy Market (NEM) data provided by the Australian Energy Market Operator (AEMO). It features:

• access to the NEMDE dataset as well as the predispatch, unit-scada, trading-price, demand and interconnectors tables from MMSDM,
• a cache to not re-download data that already exists in ~/nem-data/data,
• adds interval-start and interval-end columns.

It is designed for use by researchers & data scientists - this tool supports my personal research work. It is not designed for production use - see NEMOSIS for a production grade package.

See A Hackers Guide to AEMO & NEM Data for more context on the data provided by AEMO.

Setup

$ pip install nemdata

Use

CLI

$ nemdata --help
Usage: nemdata [OPTIONS]

  Downloads NEM data from AEMO.

Options:
  -t, --table TEXT          Available tables: nemde, dispatch-price,
                            predispatch, unit-scada, trading-price, demand,
                            interconnectors, p5min, predispatch-sensitivities,
                            predispatch-demand.
  -s, --start TEXT          Start date (YYYY-MM or YYYY-MM-DD for NEMDE).
  -e, --end TEXT            End date (incusive) (YYYY-MM or YYYY-MM-DD for
                            NEMDE).
  --dry-run / --no-dry-run  Whether to save downloaded data to disk.
  --help                    Show this message and exit.

Download NEMDE data for the first three days in January 2018:

$ nemdata -t nemde --start 2018-01-01 --end 2018-01-03

Download trading price data from MMSDM for January to March 2018:

$ nemdata -t trading-price -s 2018-01 -e 2018-03

Python

Download trading price data from MMSDM for January to Feburary 2020:

import nemdata

data = nemdata.download(start="2020-01", end="2020-02", table="trading-price")

Load this data back into a pandas DataFrame:

data = nemdata.load()['trading-price']

At this point, data will have the trading price for all regions.

Data

Downloaded into into $HOME/nem-data/data/:

$ nemdata -t trading-price -s 2020-01 -e 2020-02
$ tree ~/nem-data
/Users/adam/nem-data
└── data
    └── trading-price
        ├── 2020-01
        │   ├── PUBLIC_DVD_TRADINGPRICE_202001010000.CSV
        │   ├── clean.csv
        │   ├── clean.parquet
        │   └── raw.zip
        └── 2020-02
            ├── PUBLIC_DVD_TRADINGPRICE_202002010000.CSV
            ├── clean.csv
            ├── clean.parquet
            └── raw.zip

A few things happen during data processing:

• rows of the raw CSV are removed to create a rectangular, single table CSV,
• interval-start and interval-end timezone aware datetime columns are added,
• when using nemdata.loader.loader for trading-price, all data is resampled to a 5 minute frequency (both before and after the 30 to 5 minute settlement interval change).