johkruse/xai-for-dfds

Explainable Machine Learning for analysing deterministic grid frequency deviations

Code accompanying the mansucript "Exploring deterministic frequency deviations with explainable AI". Preprint: https://arxiv.org/abs/2106.09538

Install

The code is written in Python (tested with python 3.7). To install the required dependencies execute the following commands:

python3.7 -m venv ./venv
source ./venv/bin/activate
pip install -r requirements.txt

Usage

The scripts folder contains scripts to create the paper results and notebooks contains a notebook to reproduce the paper figures. The scripts contain a pipeline of 2 different stages:

• 1_train_test_split.py: Split data set into train and test set and save data in a version folder.
• 2_model_fit.py: Fit the XGBoost model, optimize hyper-parameters and calculate SHAP values.

Data to run the code

We use the input data (features) and output data (targets) from this zenodo repository. To run the code, copy input_actual.h5,input_forecast.h5 and outputs.h5 into a folder ./data/CE/ in the repository folder.