# kwuking/TimeMixer

[ICLR 2024] Official implementation of "TimeMixer: Decomposable Multiscale Mixing for Time Series Forecasting"

Repository: https://github.com/kwuking/TimeMixer
Canonical: https://ross.abutalabs.com/products/timemixer
Homepage: https://openreview.net/pdf?id=7oLshfEIC2
Language: Python
License: Apache-2.0
License Family: permissive
Topics: deep-learning, machine-learning, time-series, time-series-forecasting, time-series-analysis
Last push: 2025-10-05T00:30:08+00:00

## Health v2 (maintenance only)
Score: 46/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 45, release rhythm 35, longevity 65
- inputs: {"age_days": 911, "days_push": 333, "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 1981, forks 238 (observed 2026-08-28T04:06:01.815786+00:00)

## What it is
Official PyTorch implementation of TimeMixer, an ICLR 2024 model for time series forecasting using decomposable multiscale mixing. It has since expanded to support long/short-term forecasting, anomaly detection, imputation, and classification, and is integrated into PyPOTS and NeuralForecast.

## Use cases
- forecast future values of time series
- run long-term time series forecasting experiments
- detect anomalies in time series data
- impute missing values in time series
- classify time series
- reproduce ICLR 2024 TimeMixer paper results

## When to choose
- you need a state-of-the-art deep learning model for time series forecasting
- you want a research reference implementation with paper-backed results
- you need multiple time series tasks (forecasting, imputation, anomaly detection, classification) in one model family

## When to avoid
- you need a simple statistical forecasting method like ARIMA or Prophet
- you want a production-ready turnkey forecasting service rather than research code
- your data is not sequential/time-indexed

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning
- domain: machine-learning, data-science, time-series
- platform: python
- tags: time-series-forecasting, iclr-2024, research-code, multiscale-mixing, pypots

## Member repositories
- kwuking/TimeMixer (main) score 46

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:01.815786+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 2026-08-30T03:03:58.545351+00:00, confidence not recorded.
  - readme: https://github.com/kwuking/TimeMixer (fetched 2026-08-28T04:06:01.815786+00:00, sha cff8689c3fad)
- Data as of 2026-08-30T08:39:29.467469+00:00.