LLaTiSA: Towards Difficulty-Stratified Time Series Reasoning from Visual Perception to Semantics

TL;DR

LLaTiSA introduces a hierarchical time series reasoning dataset and a novel model that enhances temporal perception in vision-language models, addressing challenges in understanding complex time series data.

Contribution

The paper formalizes Time Series Reasoning with a four-level taxonomy, creates HiTSR dataset, and proposes LLaTiSA, a model with curriculum fine-tuning for improved reasoning and generalization.

Findings

LLaTiSA outperforms existing models on diverse TSR tasks.

The hierarchical dataset enables better evaluation of reasoning complexity.

The model demonstrates strong out-of-distribution generalization.

Abstract

Comprehensive understanding of time series remains a significant challenge for Large Language Models (LLMs). Current research is hindered by fragmented task definitions and benchmarks with inherent ambiguities, precluding rigorous evaluation and the development of unified Time Series Reasoning Models(TSRMs). To bridge this gap, we formalize Time Series Reasoning (TSR) via a four-level taxonomy of increasing cognitive complexity. We introduce HiTSR, a hierarchical time series reasoning dataset comprising 83k samples with diverse task combinations and verified Chain-of-Thought (CoT) trajectories. Leveraging HiTSR, we propose LLaTiSA, a strong TSRM that integrates visualized patterns with precision-calibrated numerical tables to enhance the temporal perception of Vision-Language Models (VLMs). Through a multi-stage curriculum fine-tuning strategy, LLaTiSA achieves superior performance and exhibits robust out-of-distribution generalization across diverse TSR tasks and real-world scenarios. Our code is available at https://github.com/RainingNovember/LLaTiSA.