Protein attributes contribute to halo-stability, bioinformatics approach

Aquatic Biosystems

Table 3 Comparison of percentage of correctness, wrongness, performance evaluation (T & F), and mean correct and incorrect in various decision tree models in datasets with and without feature selection for halo-tolerant and halo-stable protein groups (T/F groups)

Different decision tree models	% Correctness		% Wrongness		Performance evaluation (T)		Performance evaluation (F)		The most important feature (protein attribute) in build the decision tree
	Without feature selection	With feature selection	Without feature selection	With feature selection	Without feature selection	With feature selection	Without feature selection	With feature selection
C5.0	99.31	99.31	0.69	0.69	0.053	0.053	2.833	2.833	Count of hydrogen
C5.0 (10-fold val.)	99.31	99.31	0.69	0.69	-	0.053	-	2.833	Count of hydrogen
CR&T	99.31	98.97	0.61	1.03	-	0.057	-	2.725	Count of hydrogen
QUEST	98.97	99.31	1.03	0.69	-	0.053	-	2.833	Frequency of Leu - Leu
CHAID	99.66	99.31	0.34	0.69	-	0.053	-	2.833	Frequency of oxygen
Exhaustive CHAID	99.66	99.31	0.34	0.69	-	0.053	-	2.833	Frequency of oxygen