Search papers, labs, and topics across Lattice.
This paper introduces Context-Aware ARLLM (CAARL), a novel LLM-based forecasting approach for co-evolving time series. CAARL decomposes time series into autoregressive segments, constructs a temporal dependency graph, and serializes it into a narrative for LLM processing, enabling chain-of-thought reasoning. Experiments on real-world datasets demonstrate that CAARL achieves competitive accuracy while providing interpretable reasoning traces.
Unlock the black box of time series forecasting: CAARL uses LLMs to generate interpretable narratives that explain *why* predictions change.
In this paper we investigate forecasting coevolving time series that feature intricate dependencies and nonstationary dynamics by using an LLM Large Language Models approach We propose a novel modeling approach named ContextAware ARLLM CAARL that provides an interpretable framework to decode the contextual dynamics influencing changes in coevolving series CAARL decomposes time series into autoregressive segments constructs a temporal dependency graph and serializes this graph into a narrative to allow processing by LLM This design yields a chainofthoughtlike reasoning path where intermediate steps capture contextual dynamics and guide forecasts in a transparent manner By linking prediction to explicit reasoning traces CAARL enhances interpretability while maintaining accuracy Experiments on realworld datasets validate its effectiveness positioning CAARL as a competitive and interpretable alternative to stateoftheart forecasting methods
