Figure 7
From: Factor analysis for gene regulatory networks and transcription factor activity profiles
Reconstruction of the factor loadings matrix for the Hemoglobin data. Mean square errors (MSEs) for (a) and (c) the factor loadings matrix Λ, and (b) and (d) the factors matrix F. The positions of the zero entries in the loadings matrix are not given a priori. FA stands for the output of a given FA algorithm. On this output, a number of factor rotation methods (varimax (V), quartimax (Q), equamax (E), tanh (T) and procrustes (P)) are evaluated based on the MSE. (c) and (d) show the performance of algorithms F and W under different priors regarding the loadings matrix (for further details see section Hemoglobin dataset).