Day-ahead holiday load forecast based on pattern sequence similarity and random forest

Abstract

To solve the holiday load forecasting, a novel day-ahead holiday load forecast is proposed by means of pattern sequence similarity and random forest. The prediction for holiday load can be splitted into daily per-unit curve and daily power external value. The prediction for daily per-unit curve is conducted by pattern sequence similarity while daily power external value is predicted by random forest. Then, the above two prediction results synthesis the holiday load with segment correction. It can be found that this methodology is suitable in holiday STLF problem.