An efficient $k$-means-type algorithm for clustering datasets with incomplete records