Improving Voting Feature Intervals for Spatial Prediction of Landslides

Table 1

Formulas of quantitative indices used for validation of the models.

No.		Quantitative indices	Formulas
1		Positive predictive value (%)		[1]
2		Negative predictive value (%)		[2]
3		Sensitivity (%)		[3]
4		Specificity (%)		[4]
5		Accuracy (%)		[5]
6		Kappa		[6]
7		Root mean square error		[7]
8		Area under the ROC curve		[8]

TP, TN, FP, and FN are considered the percentage of pixels classified correctly and incorrectly as landslide and nonlandslide classes; m is the total number of instances in the datasets; and are predicted and actual values of outputs; and are expected agreements and the percentage of samples predicted correctly for landslide or nonlandslide classes; N and P are the total number of landslide and nonlandslide classes, respectively [71].