Figure 2.

Logistic regression results showing the likelihood that a penalty score would result in successful MR. The purple line describing the distribution was fitted using a sigmoid model. The coefficient and intercept were determined by the ‘LogisticRegression’ module in sklearn (http://www.scikit-learn.org). (a) The scatter points represent the 2009 raw data points, where the x value corresponds to the total penalty score and the y value is set to 1 or 0 to indicate success or failure in MR. (b) The histogram represents the proportion of success/failure for bin sizes of 1. The figure has been truncated to show the results up to a penalty score of 13; however, the sigmoid model was calculated from data sets with penalty scores of up to 26.