Research map / Time series / Causal and generative time series
Probabilistic and conformal forecasting
Produce calibrated probabilistic or conformal forecasts for series, including hierarchical and non-stationary data.
121 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Causal and generative time series, keyword cluster “conformal · causal discovery · predictions”
State History Benchmarks Tried Checklist Preprints Who Export
Where it stands
Few recent reports None of the tracked benchmarks has a report from 2025–26.
29 papers in 2023–24 and 47 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 18.7% to 20.5%.
No longer reported by papers here: traffic (CRPS-Sum, last in 2024).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Online ARIMA Algorithms for Time Series Prediction (AAAI 2016 · 196 citations)
- 2017 · Non-Stationary Spectral Kernels (NeurIPS 2017 · 42 citations)
- 2018 · Bayesian Alignments of Warped Multi-Output Gaussian Processes (NeurIPS 2018 · 11 citations)
- 2019 · High-dimensional multivariate forecasting with low-rank Gaussian Copula Processes (NeurIPS 2019 · 38 citations)
- 2020 · Normalizing Kalman Filters for Multivariate Time Series Analysis (NeurIPS 2020 · 64 citations)
- 2021 · Temporal Latent Auto-Encoder: A Method for Probabilistic Multivariate Time Series Forecasting (AAAI 2021 · 70 citations)
- 2022 · Adaptive Conformal Predictions for Time Series (ICML 2022 · 16 citations)
- 2023 · Dish-TS: A General Paradigm for Alleviating Distribution Shift in Time Series Forecasting (AAAI 2023 · 119 citations)
- 2024 · Conformal Prediction Regions for Time Series Using Linear Complementarity Programming (AAAI 2024 · 20 citations)
- 2025 · Risk and cross validation in ridge regression with correlated samples (ICML 2025 · 2 citations)
- 2026 · Ellipsoidal Time Series Forecasting (ICML 2026 · 0 citations)
Earliest papers here
- Online ARIMA Algorithms for Time Series Prediction (AAAI 2016 · 196 citations)
- Bayesian Intermittent Demand Forecasting for Large Inventories (NeurIPS 2016 · 75 citations)
- Predicting Spatio-Temporal Propagation of Seasonal Influenza Using Variational Gaussian Process Regression (AAAI 2016 · 55 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
traffic · CRPS-Sum (lower is better)
4 papers, 2019–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 0.078 | High-dimensional multivariate forecasting with low-rank Gaussian Copula Processes (NeurIPS 2019 · 38 citations) |
| 2021 | 0.069 | Temporal Latent Auto-Encoder: A Method for Probabilistic Multivariate Time Series Forecasting (AAAI 2021 · 70 citations) |
| 2024 | 0.0649 | Multivariate Probabilistic Time Series Forecasting with Correlated Errors (NeurIPS 2024 · 2 citations) |
What has been tried
81 papers here report 728 measured changes (a component added, removed or swapped, compared with a baseline): helped 67%, no clear effect 8%, hurt 19%, mixed 6%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| increase |D_cal,1| from 100 to 400 | F16 altitude validation · prediction-region width | Helped | Conformal Prediction Regions for Time Series Using Linear Complementarity Programming (AAAI 2024 · 20 citations) | |
| Add the GAN component to DT-VAE | County-DH INC · DTP score | 0.8115±0.00055 -> 0.3717±0.00017 | Helped | Cumulative Difference Learning VAE for Time-Series with Temporally Correlated Inflow-Outflow (AAAI 2024 · 8 citations) |
| Use CopulaCPTS instead of Copula | Particle Sim (σ=.01), 90% · coverage | 86.9 -> 91.3 | Helped | Copula Conformal prediction for multi-step time series prediction (ICLR 2024 · 4 citations) |
| add retrieval instead of no retrieval | Synthetic AR patterns, occurrences 1 · MSE | 0.2590 -> 0.2209 | Helped | Retrieval Augmented Time Series Forecasting (ICML 2025 · 2 citations) |
| Inject state prediction errors up to 50% instead of using ground-truth states | Bouncing Ball obs. · coverage | 90.31 \plusminus 1.27 -> 90.13 \plusminus 1.40 | No clear effect | Conformal Prediction for Time-series Forecasting with Change Points (NeurIPS 2025 · 2 citations) |
| Add local covariance ellipsoids to the global-covariance method | Multivariate time-series prediction · average prediction-set size | up to 25% reduction | Helped | Conformal prediction for multi-dimensional time series by ellipsoidal sets (ICML 2024 · 2 citations) |
| Add a low-pass filter to FITS | Time series forecasting · performance | Hurt | An Analysis of Linear Time Series Forecasting Models (ICML 2024 · 2 citations) | |
| add adaptive weight adjustment to semantic-space scores (CT-MS -> CT-SSF) | Synthetic data, miscoverage risks 0.05–0.3 · Width | null | Helped | Conformalized Time Series with Semantic Features (NeurIPS 2024 · 2 citations) |
| Increase the maximal number of maintained experts, K_max | Covertype · classification performance | Helped | Mixture of Online and Offline Experts for Non-Stationary Time Series (AAAI 2025 · 1 citation) | |
| add graph message passing and node embeddings | GPVAR-RNN · Winkler | 2.04{ ±0.00} -> 1.67{ ±0.00} | Helped | Relational Conformal Prediction for Correlated Time Series (ICML 2025 · 1 citation) |
| use approximate FE instead of FBE | Synthetic MTS · time (s) | 375±772 -> 118±178 | Helped | ChronoEpilogi: Scalable Time Series Selection with Multiple Solutions (NeurIPS 2024 · 1 citation) |
| Use Relaxed-WTA instead of annealed WTA | Solar · Distortion | 305.5 -> 280.03 | Helped | Winner-takes-all for Multivariate Probabilistic Time Series Forecasting (ICML 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 23 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Traffic · 7 of 23 papers
- Weather · 7 of 23 papers
- ETTh2 · 7 of 23 papers
- ETTm1 · 7 of 23 papers
- ETTh1 · 6 of 23 papers
- ETTm2 · 6 of 23 papers
- Solar · 5 of 23 papers
- Electricity · 4 of 23 papers
Baselines
- PatchTST · 4 of 23 papers
- ACI · 4 of 23 papers
- Autoformer · 3 of 23 papers
- iTransformer · 3 of 23 papers
- SPCI · 3 of 23 papers
- HopCPT · 3 of 23 papers
- OGD · 2 of 23 papers
- TimeGrad · 2 of 23 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 76 papers; from OpenAlex and the papers' affiliations).
- University of California, San Diego · 4 papers
- Massachusetts Institute of Technology · 4 papers
- Tsinghua University · 3 papers
- Georgia Institute of Technology · 3 papers
- Google · 3 papers
- Chinese Academy of Sciences · 2 papers
- University of Oxford · 2 papers
- University of Southern California · 2 papers
- University of Pennsylvania · 2 papers
- Harvard University · 2 papers
Use it
Download BibTeX of 39 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- State-space and latent dynamics forecasting (94 papers)
- Dynamical system reconstruction (90 papers)
- Robust forecasting and evaluation (91 papers)
- Generative temporal point processes (53 papers)
Elsewhere in the topic
- Irregular series and imputation (59 papers)
- Diffusion models for time series (55 papers)
- Neural differential equations (52 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).