# decisionintelligence/TFB

[PVLDB 2024 Best Paper Nomination] TFB: Towards Comprehensive and Fair Benchmarking of Time Series Forecasting Methods

Repository: https://github.com/decisionintelligence/TFB
Canonical: https://ross.abutalabs.com/products/tfb
Homepage: https://www.vldb.org/pvldb/vol17/p2363-hu.pdf
Language: Shell
License: MIT
License Family: permissive
Topics: benchmarking, time-series, time-series-forecasting, deep-learning, mechine-learning, statistical-learning
Last push: 2026-07-31T04:51:22+00:00

## Health v2 (maintenance only)
Score: 68/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 95, release rhythm 35, longevity 63
- inputs: {"age_days": 894, "days_push": 33, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1734, forks 123 (observed 2026-08-28T04:05:28.962648+00:00)

## What it is
TFB is a comprehensive and fair benchmarking framework for time series forecasting methods, covering deep learning, machine learning, and statistical models. It provides unified datasets, pipelines, and evaluation protocols, published as a PVLDB 2024 paper.

## Use cases
- benchmark time series forecasting models fairly
- compare deep learning vs statistical forecasting methods
- evaluate multivariate time series forecasting on standard datasets
- compute time series characteristics like trend and seasonality
- reproduce forecasting experiment results with unified hyperparameters

## When to choose
- you need a standardized, fair comparison of forecasting algorithms
- you want ready-to-use pipelines across 27+ multivariate time series datasets
- you are doing academic research on time series forecasting

## When to avoid
- you need production forecasting deployment rather than benchmarking
- your task is time series classification or anomaly detection instead of forecasting

## Facets
- artifact type: framework
- maturity: active
- function: benchmarking, machine-learning, deep-learning, data-science
- domain: time-series, machine-learning, data-science
- platform: python, cross-platform
- tags: time-series-forecasting, benchmark, pytorch, evaluation, pvldb, research, linux

## Member repositories
- decisionintelligence/TFB (main) score 68

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:28.962648+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:31:04.369691+00:00, confidence not recorded.
  - readme: https://github.com/decisionintelligence/TFB (fetched 2026-08-28T04:05:28.962648+00:00, sha f5e5eb27c54d)
- Data as of 2026-08-30T08:39:29.467469+00:00.