In PROC HPCLUS, which initialization method assigns initial cluster seeds by selecting observations that are maximally separated from each other?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because INIT=MAXDIST in PROC HPCLUS selects initial cluster seeds by choosing observations that maximize the minimum distance between seeds, producing a spread-out initialization that often leads to better convergence. A (INIT=RANDOM) selects seeds randomly from the data.
Full explanation below image
Full Explanation
B is correct because INIT=MAXDIST in PROC HPCLUS selects initial cluster seeds by choosing observations that maximize the minimum distance between seeds, producing a spread-out initialization that often leads to better convergence. A (INIT=RANDOM) selects seeds randomly from the data. C (INIT=KMEDIANS) initializes using medians, not maximum distance. D (INIT=FIRSTK) uses the first k observations as seeds.