Fig. 1From: Performance of Firth-and logF-type penalized methods in risk prediction for small or sparse binary dataPerformance of the methods was assessed using calibration slope and compared. Results were summarized over the number of simulations for which convergence is achieved. The maximum failure rate of convergence for RIDGE, out of 1000 simulations, is 8.4% when EPV=2. The values outside the whisker were not plotted to make the plot readable. The horizontal dash line is the median calibration slope for MLE and the solide line is the optimal valueBack to article page