Table 6.

Classification rules for $\mathit{Return}=\text{Yes}$ over TSCHQ and NumDays with the following parameter settings: support lower boundary $=0.01$, maximum rule length $=2$, maximum number of child rules $=800$, minimum gain in precision $=0.01$, p-value $<0.05$ (without correction for multiple testing).

Classification rules		Support	Precision	Lift
(1)	ageAtRegistration $>68$ & sex $=1$	0.01	1	1.81
(2)	ageAtRegistration $>67$ & sex $=1$	0.01	1	1.81
(3)	ageAtRegistration $>67$ & onsetrelation $=3$	0.01	1	1.81
(4)	ageAtRegistration $>65$ & familyHistory $=1$	0.01	1	1.81
(5)	ageAtRegistration $>68$ & variability $=1$	0.02	0.92	1.66
(6)	ageAtRegistration $>68$ & ageAtOnset $\le 68$	0.01	0.92	1.66
(7)	NumDays $>9$ & ageAtRegistration $\le 27$	0.02	1	1.81
(8)	NumDays $>3$ & onsetrelation $=4$	0.02	1	1.81
(9)	NumDays $>9$ & ageAtOnset $>58$	0.01	1	1.81
(10)	NumDays $>9$ & onsetrelation $=4$	0.01	1	1.81
(11)	NumDays $>9$ & NumDays $\le 10$	0.03	0.95	1.72
(12)	NumDays $>2$ & ageAtRegistration $>68$	0.02	0.93	1.68
(13)	NumDays $>5$ & familyHistory $=1$	0.1	0.91	1.64
(14)	NumDays $>9$ & familyHistory $=1$	0.07	0.9	1.62
(15)	NumDays $>20$ & onsetrelation $=3$	0.05	0.88	1.59