Application of Multiple Unsupervised Models to Validate Clusters Robustness in Characterizing Smallholder Dairy Farmers

Table 11

Proportion of variances accounted for by cluster of production in Tanzania.

	Fitted model	Total variance	Residual; Variance	-2log likelihood	P value	Variance accounted for by cluster
K-means		Milk yield
	Model with cluster	1.076	0.0027	-2981	<0.00001	71%
	Model without cluster		0.771	2584.2
		Milk sales
	Model with cluster	1.09	0.018	-1084.3	<0.00001	65%
	Model without cluster		0.723	2520
SOM		Milk yield
	Model with cluster	1.076	0.294	1633	<0.00001	44%
	Model without cluster		0.771	2584.2
		Milk sales
	Model with cluster	1.09	0.228	1381.6	<0.00001	45%
	Model without cluster		0.723	2520.2

indicates data scaled to have unit variance and mean of zero.