Skip to content

pypsa-meets-earth/demand-comp-cntry

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

demand-comp-cntry

A competition for forecasting electricity demand at the country-level using a standard backtesting framework

Introduction

This repo makes scripts available for downloading and compiling country-level data to be used in electricity demand forecasting at the country level. The goal of this repo is to encourage collaborative and competitive efforts towards the use of machine learning backtesting frameworks for forecasting electricity consumption and to use trained models to predict future consumption at the country-level. The results of such models can be used directly as input to the pypsa-africa repository and other similar modeling efforts.

Technical Background

Before getting started, it is recommended that users of and contributors to this repository have some background on backtesting, cross-validation, and probabilistic forecasting. Here are a few links to get started:

Setup

  • Set path to project root:

    cd <PATH>/demand-comp-cntry
    
  • Install conda environment:

    conda env create -f envs/environment.yml
    
  • Activate environment:

    conda activate demand-comp
    
  • Download data (basic understanding of snakemake purpose: here):

    snakemake --cores 1 download_data
    
  • Ensure you have World energy statistics (Edition 2020) data, filename iea_wes_2020-68578195-en.zip, in directory <PROJECT_ROOT>/data/raw/iea_wes_2020-68578195-en.zip (currently manual download)

  • Ensure you have World energy balances (Edition 2020) data, filename iea_web_2020-cde01922-en.zip, in directory <PROJECT_ROOT>/data/raw/iea_web_2020-cde01922-en.zip and unzip these (currently manual download)

  • Run test script using ARIMA models:

    snakemake --cores 1 run_arima
    
  • Analyse the outputs in demand-comp-cntry/out

About

A competition for forecasting electricity demand at the country-level using a standard backtesting framework

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages