Extended Data Fig. 1: Computational experiment for growth parameters of logistic and Gompertz growth models.
The vertical coordinates of each dot show the divergence in G (max growth rate) and horizontal – in ∆t (the duration of transition) for a single computational experiment. The divergence is the ratio between the ‘true’ growth parameter of a computer-generated logistic (a) and Gompertz (b) curve and the same parameter estimated by fitting the other model to the same data (Methods). The dashed line corresponds to equal divergence of G and ∆t. Experiments differ by the degree of curve ‘maturity’ (represented by different colors), that is to which extent the generated data approach the asymptote L of the curve (Methods). Large triangles show the relative differences between the ‘true’ curve and the fitted growth model. Dots show the divergence between the original curve with random noise added (up to 5% above or below the respective original value, uniformly distributed) and the fitted model (Methods). The data illustrate that estimates of the growth parameters converge across the two models with more complete data (high maturity), but that the max growth rate G becomes robust across the two models at lower levels of maturity than the duration of transition ∆t. See Supplementary Note 3 for more discussion on growth metrics.
