Table 2 Mean absolute error (MAE) results for the two user groups MS and BeyMS of different mainstreaminess and a selection of standard recommendation algorithms. A one-tailed Mann–Whitney-U test (\(\alpha = 0.0001\)) provides significant evidence, indicated by ^∗∗∗, that all algorithms perform worse on BeyMS than on MS in terms of MAE. Furthermore, NMF (as shown in bold) outperforms the other three approaches UserItemAvg, UserKNN and UserKNNAvg

From: Support the underground: characteristics of beyond-mainstream music listeners