Model 1.3
- Citation Author(s):
-
Konstantinos Theodorakos
- Submitted by:
- Konstantinos Theodorakos
- Last updated:
Abstract
Model 1.3
Gradient Linear Booster using Dart (Python xgboost library)
Monthly (kWh) consumption forecasting:
1. Twelve regression models (one per month).
2. Regression using (daily kWh) consumption extracted features.
From weekend/weekday and full series:
- Statistical: median, variance, quantiles, ...
- Time-series: autocorrelations, trends, seasonalities, ..
Regression model training: Nested cross-validation.
- Outer k-fold CV loop splits held-out test data.
- Inner k-fold CV loop splits train-validation data.
Hyperparameter optimization on the outer loop using: Dual annealing (global + local search).
The Dart XGBoost booster type vs the default (gbtree): uses drop-out ratios (of trees tou leave out) while training, inspired by Deep Learning techniques.
