This repository contains the experiments related with a new baseline model that can be used in forecasting weekly time series. This model uses the forecasts of 4 sub-models: TBATS, Theta, Dynamic Harmonic Regression ARIMA and a global Recurrent Neural Network (RNN), and optimally combine them using lasso regression.
Contributions:32 commits, 15 pushes, 1 branch in 1 year 7 months
arimalasso-regressionrecurrent-neural-networkstime-series
This repository contains the experiments related to a new and accurate tree-based global forecasting algorithm named, SETAR-Tree.
Contributions:26 commits, 27 pushes, 1 branch in 1 year 9 months