Table 1 The number of the true predictions and mean absolute error of k-medians, naïve GMM, AGMM without the outlier detection and AGMM with the outlier detection. AGMM with the outlier detection made more true predictions, and its predictions have the least mean absolute error
From: l1kdeconv: an R package for peak calling analysis with LINCS L1000 data
|  | k-medians | naïve GMM | AGMM without the outlier detection | AGMM with the outlier detection |
|---|---|---|---|---|
# of true predictions | 614 | 1040 | 1342 | 1416 |
Mean absolute error | 0.8412 | 0.4818 | 0.2797 | 0.2545 |