GEFCOM2014

This repository has the code and description of the modelling approach that I used for the Load track of the Global Energy Forecasting Competition: 2014, hosted on CrowdAnalytix.

Description of files

• final_report.md: A summary of my modelling approach as well as a log of what changed week by week
• gef2014_functions.R: Various functions used to load data, derive variables, and build models
• train_pca_model.R: Extracting principal components from the weather variables
• train_gbm_models.R: Training the quantile regression gradient boosting machine models
• predict_distribution.R: Simulating weather and predicting the distribution on the next month