Reducing Uncertainty in Dynamic Stall Measurements Through Data-Driven Clustering of Cycle-To-Cycle Variations