Understanding the Impact of Custom K-Means Initialization on Clustering Results in R