Fig. 7From: Estimating Phred scores of Illumina base calls by logistic regression and sparse modelingThe AUC of the ROC curve versus the number of features in the initial model. In the logistic model, we sequentially include one more feature starting from x 0 to x 74 in Table 1 (the number of features is shown by the x-axis), and calculated each AUC (shown by the y-axis) using the L 1-regularized methodBack to article page