This repository contains the code for the outage forecasting using Outage Watch: Early Prediction of Outages using Extreme Event Regularizer
The directory consists of :
model: This folder contains the proposed BiLSTM -MDN + Classifier model and the baseline ablation models.
- This code is a sample boilerplate with limited dataset due to constraints outlined in Section 11.
- Running the scripts will produce results with the sample dataset only.
- Certain parts of the code that are not applicable to the sample dataset have been excluded, such as the alert-based extreme event label filtering and EVL loss parameters.
- This repository contains python files for the classifier, MDN, and MTL code and their respective main.ipynb notebooks for execution.
Please contact with authors for further info.
