Global Dataset of Forecasted Hourly Electricity Demand from 2000 to 2024 Using DemandCast

The dataset included in this repository provides forecasted hourly electricity demand for multiple countries worldwide. The forecasts are generated using DemandCast, a demand forecasting framework trained on historical electricity demand, meteorological variables, and socioeconomic indicators. Hourly electricity demand forecasts are currently generated at the national level, with coverage for 184 countries. Comprehensive documentation of the modeling approach, data inputs, and generation pipeline is available in the DemandCast GitHub repository: https://github.com/open-energy-transition/demandcast. This repository contains the following files: forecasts_on_historical_period.parquetContains the model-generated hourly electricity demand forecasts over the historical period (2000–2024). These forecasts are produced by DemandCast using information available at the time of prediction. Not all countries have forecasts covering the entire historical period due to missing or insufficient input data. input_data_for_forecasting_on_historical_period.parquetProvides the full set of input features used to generate the historical forecasts, including meteorological variables and socioeconomic indicators. list_of_country_codes_not_yet_available.txtLists of country codes for which forecasts are not yet available due to missing or insufficient input data. trained_xgboost_model.jsonThe XGBoost model trained within the DemandCast framework.