TimeHetNet (Meta Dataset for Time Series with heterogeneous networks)
Introduced by Brinkmeyer et al. in Few-Shot Forecasting of Time-Series with Heterogeneous Channels
This meta-dataset is composed of previously known datasets.
This includes: It also uses a specific script to read and sample small tasks from specified sizes and lengths.
DS included here are from :
PeekDB (https://github.com/RafaelDrumond/PeekDB) Informer data-sets (https://github.com/zhouhaoyi/Informer2020) Monash (https://zenodo.org/communities/forecasting) UEA (http://www.timeseriesclassification.com/Downloads/Archives/Multivariate2018_arff.zip) CNC (https://www.kaggle.com/datasets/shasun/tool-wear-detection-in-cnc-mill/download) MINING (https://www.kaggle.com/datasets/edumagalhaes/quality-prediction-in-a-mining-process/download) Plant_Monitoring (https://www.kaggle.com/datasets/inIT-OWL/production-plant-data-for-condition-monitoring/download)
Licenses from each db differs individually.
Benchmarks
| Trend | Task | Dataset Variant | Best Model | Paper | Code |
|---|---|---|---|---|---|
|
TimeHetNet
|
TimeHetNet
|
Papers
| Paper | Code | Results | Date | Stars |
|---|
