Independently of the out-of-sample extension, however, L-IsomapSC led to more accurate predictions with respect to L-IsomapMSC and performed better than its unsupervised variant, which emerged as the second best fast Isomap algorithm according to the empirical tests