Research map / Time series / Foundation models for time series
It's TIME: Towards the Next Generation of Time Series Forecasting Benchmarks
Zhongzheng Qiao, Sheng Pan, Anni Wang, Viktoriya Zhukova, Yong Liu, Xudong Jiang, Qingsong Wen, Mingsheng Long, Ming Jin, Chenghao Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use newer model versions (TimesFM-2.5 vs 2.0/1.0, Moirai2 vs Moirai, Chronos-2 vs Chronos-bolt) | TSFMs · TIME (98 tasks) | normalized MASE and CRPS | Improved |
| rank models by task-level versus variate-level aggregation | TSFM leaderboard · TIME | MASE/CRPS ranking | Mixed |
| evaluate on stationary versus non-stationary variates | TSFMs vs Seasonal Naive · TIME pattern F7 | normalized MASE | Mixed |
| strong trend (high trend strength) variates | most TSFMs · TIME pattern F1 | normalized MASE | Improved |
| high seasonality strength variates | TSFMs · TIME pattern F3 | MASE spread between models | Mixed |
| high spectral complexity variates | TSFMs · TIME pattern F6 | inter-model disparity | No clear effect |
| newer TSFMs on stable versus unstable seasonality | Chronos-2 and TimesFM-2.5 · TIME pattern F4 | normalized MASE | Improved |
| flat conservative forecasts on high-variation series | TSFMs · TIME | MASE/CRPS | Mixed |
Closest papers by meaning
- fev-bench: A Realistic Benchmark for Time Series Forecasting (NeurIPS 2026 · similarity 0.91)
- Less is More: Unlocking Specialization of Time Series Foundation Models via Structured Pruning (NeurIPS 2025 · similarity 0.90)
- TimeBase: The Power of Minimalism in Efficient Long-term Time Series Forecasting (ICML 2025 · similarity 0.90)
- Time-MMD: Multi-Domain Multimodal Dataset for Time Series Analysis (NeurIPS 2024 · similarity 0.90)
- TS-RAG: Retrieval-Augmented Generation based Time Series Foundation Models are Stronger Zero-Shot Forecaster (NeurIPS 2025 · similarity 0.89)
- Context is Key: A Benchmark for Forecasting with Essential Textual Information (ICML 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
