. 2020 May 5;8(5):e14330. doi: 10.2196/14330

Table 5.

Agreement between elastic-net regularized generalized linear model, maximum entropy, and boosting using 5-fold cross-validation.

Technique	Wrongly agree^a, n	Correctly agree^b, n	Disagree^c, n	Gwet AC1^d,e (95% CI)
GLMNet^f vs MAXENT^g	669	5609	1353	0.68 (0.67-0.70)
GLMNet vs boosting	195	6269	1146	0.74 (0.72-0.75)
MAXENT vs boosting	224	5895	1491	0.66 (0.65-0.68)

^aThe “Wrongly Agree” column refers to the number of records misclassified by both techniques.

^bThe “Correctly Agree” column states the number of records correctly classified by both techniques.

^cThe “Disagree” column lists the number of records for which the techniques disagree in the classification.

^dAC1: agreement coefficient 1.

^eGwet AC1 represents the index of agreement between the identified techniques. Legend for AC1 is: AC1<0=disagreement; AC1 0.00-0.40=poor; AC1 0.41-0.60=discrete; AC1 0.61-0.80=good; AC1 0.81-1.00=optimal.

^fGLMNet: elastic-net regularized generalized linear model.

^gMAXENT: maximum entropy.