Identification of AIDS Disease Severity Using Genetic Algorithm - Computational Intelligence Techniques for Comparative Genomics - page 99

A
1

A
2

A
3

A
4

Classes(Two)
Positive or negative

Fig. 2 Rule consisting of attributes

= >

Fig. 3 Operator combination for the above attribute values

P Value is the number of samples in the dataset that are satisfying only the
antecedent part in the rule.
Q Value is the number of samples in the dataset that are satisfying only the
antecedent part in the rule.
Rule length calculation represents the number of attributes involved in rule
formation.
L is the maximum possible length of the rule (Number of attributes).
B is the length of the current rule.
True Positives (TP): This refers to the positive tuples that were correctly labeled
by the classi er.

True Negatives (TN): This refers to the negative tuples that were correctly
labeled by the classi er.

False Positives (FP): This refers to the negative tuples that were incorrectly
labeled as positive.
False Negatives (FN): This refers to the positive tuples that were mislabeled as
negative.
Sensitivity: It measures the ability of the method to identify the occurrence of
target class accurately.
Speci

city: It measures the ability of the method to separate the target class.
Comprehensibility: The comprehensibility of the rule R can be de ned by the

number of attributes on the left-hand side of the rule.
TP = P Q;FP=P (P Q); FN = Q (P Q).

˄ - ˄ - ˄

TN = (Number of samples in the dataset) P Q + P Q.

- - ˄

TP

Sensitivity =

TN ;

þ

TP

TN

Speci

FP ;
Comprehensibility = L B
L

city =

TN þ

1 ;
Fitness = Sensitivity * Speci

city + 0.2 * (Comprehensibility) [ 8 ].
The AIDS parameters like in CD4 cell count are Marital status, Age, Gender, etc.
By using these we developed several rules. For CD4 cell count parameter using