5.2 Performance Analysis Factors

5.2. Performance Analysis Factors
The most widely used measures to assess the performance of diagnosis the disease systems is as follows. Table 7 shows the confusion matrix containing the information about actual and predicted classifications which is used to evaluate the performance metrics. The entries in the confusion matrix have the following meaning in the context of our study: tp (true positives) is the number of cases covered by the rule that have the class predicted by the rule. fp (false positives) is the number of cases covered by the rule that have a class different from the class predicted by the rule. fn (false negatives) is the number of cases that are not covered by the rule but that have the class predicted by the rule. tn (true
To measure agreement between two rater (The expert’s analysis and data mining techniques analysis) classifying the same set of cases can be calculated by using of Kappa. Cohen’s kappa measures the agreement between two rater (e.g. an expert and a data mining technique) classifying the same set of cases. Cohen’s kappa defines the measure of agreement as the ratio of the percentage of agreement minus the chance agreement to the largest possible non chance agreement. This measure, thus, takes into account the classifications that could match merely by chance. The chance agreement actually depends upon the percentage of matches in each class, and it reduces as the number of classes increases. Using the above definition, a kappa value of 1 indicates a perfect agreement and a kappa value of 0 indicates that agreement is no better than chance
The performance of F score for the proposed GA-SM algorithm and CN2 is equal(0.78), whereas for J48 and BF tree technique it is 0.39 and 0.336.The kappa value for the GA-SM generates more reasonable level of agreement between experts’ problem-solving knowledge than CN2 , J48 and BF tree. This indicates that the GA method is effective in the discovery of experts’ subjective knowledge .Where the Number of observed agreements: 654 ( 85.16% of the observations) , Number of agreements expected by chance: 452.9 ( 58.97% of the observations), Kappa=0.638 , SE of kappa = 0.029 and 95% confidence interval: From 0.581 to 0.695. Here the strength of agreement is considered to be 'good'. For CN2, Number of observed agreements: 533 ( 69.40% of the observations) Number of agreements expected by chance: 490.0 ( 63.81% of the observations) Kappa= 0.155 , SE of kappa = 0.025 and 95% confidence interval: From 0.106 to 0.203 .Here the strength of agreement is considered to be 'poor'. For J48 ,Number of observed agreements: 567 ( 73.83% of the observations) ,Number of agreements expected by chance: 423.6 ( 55.15% of the observations) Kappa= 0.416, SE of kappa = 0.034 and 95% confidence interval: From 0.349 to 0.484.Here the strength of agreement is considered to be 'moderate'. Number of observed agreements: 565 ( 73.57% of the observations). For BF tree , Number of agreements