{"id":28729825,"url":"https://github.com/didi/heteta","last_synced_at":"2025-06-15T17:11:22.051Z","repository":{"id":62925375,"uuid":"272919482","full_name":"didi/heteta","owner":"didi","description":"HetETA: Heterogeneous Information Network Embedding for Estimating Time of Arrival","archived":false,"fork":false,"pushed_at":"2020-06-29T02:21:03.000Z","size":26730,"stargazers_count":101,"open_issues_count":4,"forks_count":43,"subscribers_count":8,"default_branch":"master","last_synced_at":"2024-04-14T13:44:08.272Z","etag":null,"topics":["estimated-time-of-arrival","graph-neural-networks","traffic-prediction"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/didi.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2020-06-17T08:28:11.000Z","updated_at":"2024-02-29T04:54:43.000Z","dependencies_parsed_at":"2022-11-09T08:00:19.592Z","dependency_job_id":null,"html_url":"https://github.com/didi/heteta","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/didi/heteta","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/didi%2Fheteta","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/didi%2Fheteta/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/didi%2Fheteta/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/didi%2Fheteta/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/didi","download_url":"https://codeload.github.com/didi/heteta/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/didi%2Fheteta/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":260016055,"owners_count":22946321,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["estimated-time-of-arrival","graph-neural-networks","traffic-prediction"],"created_at":"2025-06-15T17:11:13.358Z","updated_at":"2025-06-15T17:11:22.037Z","avatar_url":"https://github.com/didi.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"HetETA: Heterogeneous Information Network Embedding for Estimating Time of Arrival\n---------------\n\nThis is basic implementation of our KDD'20 Applied Data Science Track (Oral) paper:\n\nHuiting Hong, Yucheng Lin, Xiaoqing Yang, Zang Li, Kung Fu, Zheng Wang, Xiaohu Qie, Jieping Ye. 2020. HetETA: Heterogeneous Information Network Embedding for Estimating Time of Arrival.\n\nThe source code is based on [STGCN](https://github.com/VeritasYin/STGCN_IJCAI-18)\n\nHetETA framework            \n:-------------------------:\n![](https://github.com/didi/heteta/raw/master/figs/framework.png)\n\n\nDependencies\n------------\nThe script has been tested running under Python 2.7.5, with the following packages installed (along with their dependencies):\n\n- `argparse==1.1`\n- `numpy==1.16.5`\n- `scipy==1.2.2`\n- `networkx==2.2`\n- `tensorflow-gpu==1.13.1`\n- `yaml==5.1.2`\n\n\n\nOverview\n--------------\nHere we provide the implementation of HetETA and a toy dataset.\n\nThe folder is organised as follows:\n- `dataset/` contains:\n    - `make_sample.py` randomly generates the `toy_sample` dataset to help readers to figure out the input format;\n    - `toy_sample/` contains:\n        * `adj_gap_top5.mat` is the vehicle-trajectories based network;\n        * `adj.mat` is the multi-relational road network;\n        * `link_info.npz` is the static attributes of each road segment;\n        * `dynamic_fes.npz` is the dynamic feature (speed) of each road segment over time periods;\n        * `eta_label.npz` contains the time it takes for a vehicle to travel through a path starting form period t.\n- `codes/` contains:\n    - `data/`:\n        * `model/` is used to save the trained model;\n        * `config_*.yaml` configures the path and paramenter settings.\n    - `model/` contains the implementation of the HetETA network;\n    - `utils/` contains some tools for loading dataset;\n    - `train.py` is used to execute a full training run on the dataset.\n\n\nHow to run\n---------------\n\n```shell\ncd codes\npython -u train.py --config data/config_HetETA_toy.yaml --model_dir data/model/HetETA_toy --dataset_dir ../dataset/toy_sample \u003e\u003e multi-HetETA_toy.log\n```\n\nPlease note that the `toy_sample` dataset is not a real dataset, which is only used to provide examples of data formats, not to train models.\n\nLicense\n----------\nDidi Chuxing, Beijing, China.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdidi%2Fheteta","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fdidi%2Fheteta","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdidi%2Fheteta/lists"}