Interpretation of Load Forecasting Using Explainable Artificial Intelligence Techniques

Artificial intelligence (AI) is getting popular and has been successfully applied to many applications. However, in many cases, AI is considered as a ‘black box’ which is hard to interpret. Recently, researchers have been attempting to explain AI systems and various explainable AI techniques have been developed. In this paper, we apply explainable AI techniques to interpret the load forecasting based on machine learning method. For load forecasting, we employ XGBoost which is decision tree based gradient boosting algorithm. The XGBoost based load forecasting approach was analyzed in terms of feature importance and partial dependence plot. The experimental results show that the performance can be improved by selecting features which were found to have high importance in the SHAP analysis.