Evaluation of interpretability methods for multivariate time series forecasting experiments

Paper link: https://link.springer.com/article/10.1007/s10489-021-02662-2

Citation

@article{ozyegen2022evaluation,
  title={Evaluation of interpretability methods for multivariate time series forecasting},
  author={Ozyegen, Ozan and Ilic, Igor and Cevik, Mucahit},
  journal={Applied Intelligence},
  volume={52},
  number={5},
  pages={4727--4743},
  year={2022},
  publisher={Springer}
}

Experiments

Evaluation of local explanations is performed in three steps:
1- Train the models for the datasets
2- Generate feature importances using local explanation methods
3- Evaluate feature importances

Steps

You can easily reproduce the experiments using the Makefile.

• Train models
  • make train_models
• Evaluate local explanations
  • make eval_metrics
• Report the training results
  • make report_train

Data Sources

Electricity

• https://archive.ics.uci.edu/ml/datasets/ElectricityLoadDiagrams20112014

Rossmann

• https://www.kaggle.com/c/rossmann-store-sales

Walmart

• https://www.kaggle.com/c/walmart-recruiting-store-sales-forecasting/data

Ohio

• You must request access from the following link
• http://smarthealth.cs.ohio.edu/OhioT1DM-dataset.html

Requirements

• Python 3.7 and Tensorflow 2.2
• wandb - Weights and Biases is used for tracking the experiments
• xai.yml contains all the package dependencies