Parallelizing Dynamic Time Warping Algorithm for Hotel Occupancy Time Series Similarity Measurement

This paper addresses the challenge of computationally efficient similarity calculations between diverse hotel occupancy time series. The volume of data generated daily by the hospitality industry introduces challenges, particularly in the domain of similarity computation between distinct hotel occupancy time series. Given the significance of similarity analysis in making informed business decisions and optimizing profits, rapid and efficient computation of this metric becomes very significant for the sector’s advancement. Through a series of experiments, this study investigates the prospect of accelerating the calculation of Dynamic Time Warping (DTW) distance, a highly relevant metric in time series analysis, between different hotel occupancy time series using CUDA parallel processing, aiming to provide a practical solution for addressing the associated computational challenges.