PROC HPCLUS is run with NCLUST=5 on a 10-million-row dataset. The IMPUTE option is specified. What does IMPUTE do in PROC HPCLUS?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because PROC HPCLUS's IMPUTE option handles missing values by assigning incomplete observations to their nearest cluster using available (non-missing) variables, then imputing missing values with the centroid values of the assigned cluster. This preserves all observations in the clustering solution.
Full explanation below image
Full Explanation
B is correct because PROC HPCLUS's IMPUTE option handles missing values by assigning incomplete observations to their nearest cluster using available (non-missing) variables, then imputing missing values with the centroid values of the assigned cluster. This preserves all observations in the clustering solution. A (mean imputation before clustering) is different and does not use cluster centroids. C (complete case analysis) is the default behavior without IMPUTE. D is wrong; PROC HPCLUS does not use PROC MI internally.