##IRDM Group Project -- Global Energy Forecasting Competition 2012: Load Forecasting

###src Directory: all code are in root directory(python and R) filename start with 'old' are for models which are not discussed in our report.

###data Directory: all files in this directory are downloaded from https://www.kaggle.com/c/global-energy-forecasting-competition-2012-load-forecasting/data

###features Directory: csv files containing data after feature engineering are saved in this directory. load_features.csv has only features generated from Load_history.csv GP_temppred.csv has only features generated from temperature.csv loadtemp_features.csv is the combination of load_features.csv and GP_temppered.csv file loadtemp_features_withsmoothtemp.csv is by add smoothed tempeartures to loadtemp_features.csv file(part of smoothed tempteratures are acquired from https://github.com/jamesrobertlloyd/gp-structure-search)

###output Directory: ann_keras_fit.csv is the result for Neural Network, and result is used in the report. svm_output.csv is the result for SVM, and result is used in the report. gbm_output3.csv is the result for Gradient Boosting, and result is used in the report. arima_fit.csv it the result for ARIMA(this is only done for backcast), result is not good and not used in the report.