Optimizing Window Size Determination for Improved Forecast Accuracy Through Pattern Sequence Similarity Detection

This research paper introduces an innovative approach to electricity demand forecasting, offering valuable insights for electricity departments seeking efficient load balancing and precautionary measures. The Novel Approach enhances the existing algorithm of pattern sequence-based forecasting (PSF) by optimizing it by identifying an ideal window size determined via lag analysis. The procedure involves detecting patterns in electrical demand data using a pattern sequence similarity detection technique and subsequently determining the ideal window size to improve prediction accuracy. Evaluation of two widely used electric power consumption datasets demonstrates the superior performance of the proposed model over traditional linear prediction and ARIMA models. The model's scalability for real-time big data processing is achieved by deploying it on Apache Spark.