TABLE 3.

Confusion matrix when predicting the 20% unseen testing dataset using the model obtained through the 80% training set, where (A) used the optimal tree model and (B) the reduced tree model.

Prediction	A reference			B reference
	Good	Moderate	Poor – very poor	Good	Moderate	Poor – very poor
Good	46	7	2	50	5	0
Moderate	10	15	4	14	12	3
Poor – very poor	2	4	9	3	4	8

Note: The matrix displays the number of cases of each nutritional status.