# liyaguang/DCRNN

Implementation of Diffusion Convolutional Recurrent Neural Network in Tensorflow

Repository: https://github.com/liyaguang/DCRNN
Canonical: https://ross.abutalabs.com/products/dcrnn
Language: Python
License: MIT
License Family: permissive
Topics: traffic-data, deep-learning-graphs, time-series, iclr2018, spatiotemporal-forecasting
Last push: 2024-12-09T17:40:10+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3191, "days_push": 632, "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 1439, forks 423 (observed 2026-08-28T04:04:44.119676+00:00)

## What it is
A TensorFlow implementation of the Diffusion Convolutional Recurrent Neural Network (DCRNN) from the ICLR 2018 paper on data-driven traffic forecasting. It models road networks as graphs and forecasts traffic time series on benchmark datasets like METR-LA and PEMS-BAY.

## Use cases
- forecast traffic speed on highway sensor networks
- reproduce DCRNN results from the ICLR 2018 paper
- train spatiotemporal graph neural networks on METR-LA or PEMS-BAY
- benchmark traffic prediction models
- learn graph-based time series forecasting with TensorFlow

## When to choose
- you need a reference implementation of DCRNN for research or benchmarking
- you want to experiment with graph recurrent networks for traffic or sensor forecasting
- you are working with the METR-LA or PEMS-BAY datasets

## When to avoid
- you need production-ready traffic forecasting with modern framework support
- you want PyTorch or newer TensorFlow 2.x compatibility out of the box
- your sensor network is not the Los Angeles road graph, since adjacency is precomputed for specific sensor ids

## Facets
- artifact type: library
- maturity: maintenance
- function: deep-learning, machine-learning, data-science, streaming
- domain: deep-learning, machine-learning, time-series
- platform: python
- tags: tensorflow, graph-neural-networks, traffic-forecasting, spatiotemporal, iclr-2018, research-code, metr-la, pems-bay, transportation, linux, macos, gpu

## Member repositories
- liyaguang/DCRNN (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:44.119676+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 2026-08-30T04:36:36.632665+00:00, confidence not recorded.
  - readme: https://github.com/liyaguang/DCRNN (fetched 2026-08-28T04:04:44.119676+00:00, sha 77f58aba8689)
- Data as of 2026-08-30T08:39:29.467469+00:00.