Table 2. Comparison matrices showing model superiorities over one another, for both data sets.

Superiority is defined as the proportion of one model’s calibration Pareto front (rows) non-dominated by the other’s (columns, see S11 Fig). Each model is represented by a Pareto front representing the pooled Pareto-optimal solutions arising from three independent calibrations. Values in parentheses represent Pareto front sizes.

T cell dataset
		Non-dominated by
		Brownian	Lévy	HomoCRW	HeteroCRW	IHomoCRW	IHeteroCRW
		(65)	(196)	(248)	(152)	(298)	(104)
Brownian	(65)	-	0	0	0	0	0
Lévy	(196)	100	-	1.0	0.5	0.5	1.0
HomoCRW	(248)	100	100	-	4.4	27.0	2.4
HeteroCRW	(152)	100	100	98.0	-	98.0	9.9
IHomoCRW	(298)	100	100	84.9	6.4	-	5.0
IHeteroCRW	(104)	100	100	99.0	95.2	99.0	-
Neutrophil dataset
		Non-dominated by
		Brownian	Lévy	HomoCRW	HeteroCRW	IHomoCRW	IHeteroCRW
		(124)	(343)	(458)	(239)	(386)	(139)
Brownian	(124)	-	7.3	0	0	0	0
Lévy	(343)	100	-	72.3	31.5	43.7	45.2
HomoCRW	(458)	100	30.3	-	7.2	8.5	17.9
HeteroCRW	(239)	100	82.0	94.1	-	88.3	67.4
IHomoCRW	(386)	100	56.7	97.7	39.6	-	42.0
IHeteroCRW	(139)	100	75.5	100	50.4	91.4	-