Table 1 Predictive power for farm financial high and low well-being.

Accuracy	wellbeing	Sgb	mb	Mb int	glm	rf	nn	gbm
Chile	High	0.65	0.675		0.642	0.733	0.575	0.683
Chile	Low	0.833	0.833		0.833	0.842	0.833	0.817
Chile & Tunisia	High	0.71	0.693	0.685	0.618	0.734	0.556	0.705
Chile & Tunisia	Low	0.809	0.809		0.822	0.817	0.822	0.793
Tunisia	High	0.595	0.579		0.545	0.645	0.529	0.57
Tunisia	Low	0.76	0.744		0.736	0.769	0.529	0.727
AUC	wellbeing	Sgb	mb	Mb int	glm	rf	nn	gbm
Chile	High	0.655	0.717		0.687	0.757	0.603	0.727
Chile	Low	0.830	0.802		0.676	0.837	0.642	0.773
Chile & Tunisia	High	0.733	0.723	0.731	0.627	0.796	0.619	0.758
Chile & Tunisia	Low	0.763	0.733		0.637	0.746	0.721	0.735
Tunisia	High	0.663	0.661		0.537	0.710	0.616	0.658
Tunisia	Low	0.596	0.562		0.579	0.614	0.492	0.579

The accuracy and Area Under the Curve (AUC) of all fitted models was evaluated on the test data from Chile and Tunisia. For corresponding receiver operator curves (see Supplementary Fig. 2). For abbreviation explanation, see Methods - Choice of predictive models for data analysis.

Quick links

Search