Skip to content

Latest commit

 

History

History
20 lines (13 loc) · 1.35 KB

README.md

File metadata and controls

20 lines (13 loc) · 1.35 KB
Raw

GB Power Demand Forecasting using Prophet

Description

This repository provides tools for forecasting the power demand of Great Britain (GB) using Facebook's Prophet model. The repository includes Python functions and Jupyter notebooks that illustrate how to apply Prophet for daily and monthly forecasting, quantifying the effect of weather and other regressors (e.g., GDP), and using HyperOpt for tuning hyperparameters.

Key Features:

  • Daily and Monthly Power Demand Models: Separate models for short-term (daily) and long-term (monthly) forecasting.
  • Weather and GDP Regressors: Integration of weather data and GDP as external regressors to improve forecasting accuracy.
  • Hyperparameter Tuning with HyperOpt: Automated tuning of Prophet's hyperparameters for optimized performance.
  • Comprehensive Notebooks: Detailed notebooks explaining the internals of Prophet, including mathematical foundations.
  • Presentation: A presentation explaining all the key takeaways and also the mathematical foundations behind Prophet.

Usage

  • In the modelling folder you can find notebooks that take you step-by-step through the modelling process.
  • The Powerpoint presentation presents the mathematical foundations in a clear way and also the main takeaways and explanation of different variables/parameters.