Fit performances of growth models for the breast data.

<p>Models were ranked in ascending order of the <i>RMSE</i>, defined by expression (13). For each metric, indicated are the mean value (among all animals) and in parenthesis the minimal and maximal values (not reported for <i>AICc</i> as they were redundant with the range of <i>AIC</i>). When reported, value inside brackets is the rank of the model for the underlying metric. The model ranking first is highlighted in bold. For animal <i>j</i>, is the minimal value of the objective that was minimized in the individual fits approach (see (8)), divided by the number of time points <i>I^j</i>, and represents the variance of the weighted residuals. <i>AIC</i> and <i>AICc</i> are defined in (11) and (12), <i>AICc_pop</i> is the <i>AICc</i> resulting from the mixed-effect estimation (see <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1003800#s2" target="_blank">Materials and Methods</a>) and <i>R</i>² is defined in (14). Values reported in the <i>p</i> column are percentages of animals for which Kolmogorov-Smirnov test for normality of residuals was not rejected at the significance level of 0.05. # = number of parameters. <i>J</i> = 34 animals.</p><p>Fit performances of growth models for the breast data.