In this paper, we present a novel method for advancing time series forecasting by representing time series data through de Bruijn Graphs (dBGs). This method harnesses the capability of dBGs to encapsulate and project future states from historical sequences, thus enhancing predictive analytics in time series.
Our approach is multi-faceted, involving:
- The enhancement of dBGs via a suite of graph substitutes, allowing to capture patterns better.
- The application of advanced graph encoding techniques, specifically struct2vec, to distill salient features from these intricate graph structures.
- The seamless integration of these extracted features into the TimesNet model to bolster short-term forecasting accuracy.
Empirical evaluations conducted on the M4 datasets illustrate that our approach not only maintains the intrinsic dynamics of the time series but also achieves notable improvements in forecasting performance across diverse datasets.
Kurban Intelligence Lab 