Boosted Ensemble Learning based on Randomized NNs for Time Series Forecasting

TL;DR

This paper introduces a boosted ensemble learning approach using randomized neural networks for complex time series forecasting, enhancing accuracy and speed by unifying forecasting tasks across ensemble members.

Contribution

The paper proposes a novel ensemble boosting method based on randomized neural networks that simplifies task unification and improves forecasting accuracy for complex time series.

Findings

Ensemble methods improve forecasting accuracy for complex time series.

The proposed methods achieve rapid training and effective pattern extraction.

Experimental results confirm the effectiveness of the boosting variants.

Abstract

Time series forecasting is a challenging problem particularly when a time series expresses multiple seasonality, nonlinear trend and varying variance. In this work, to forecast complex time series, we propose ensemble learning which is based on randomized neural networks, and boosted in three ways. These comprise ensemble learning based on residuals, corrected targets and opposed response. The latter two methods are employed to ensure similar forecasting tasks are solved by all ensemble members, which justifies the use of exactly the same base models at all stages of ensembling. Unification of the tasks for all members simplifies ensemble learning and leads to increased forecasting accuracy. This was confirmed in an experimental study involving forecasting time series with triple seasonality, in which we compare our three variants of ensemble boosting. The strong points of the proposed ensembles based on RandNNs are extremely rapid training and pattern-based time series representation, which extracts relevant information from time series.