This research further expands recent studies concerning classification and regression methodologies that utilize Latin Hypercube Sampling (LHS), as well as applying Progressive LHS (PLHS) in classification, by exploring the implementation of PLHS in the field of unsupervised machine learning. PLHS is a sampling approach characterized by the sequential accumulation of matrices on an initial Latin Hypercube Design (LHD) matrix, aiming at the progressive construction of LHD matrices and thereby optimizing space filling. The primary goal is to investigate the effectiveness of this sampling method in addressing clustering problems, particularly in comparison to a standard sampling technique such as Random Sampling (RS). For this purpose, three datasets were assessed using five widely used clustering algorithms and then evaluated with two well-established metrics and an additional one introduced specifically for this study. The findings suggest that PLHS demonstrates considerable promise, as the results generally favor its performance over that of RS.