Comparative performance analysis of different classifiers on diagnosis of erythmato-squamous diseases

The differential diagnosis of Erythemato-Squamous diseases is a challenging task due to their incomparable features. As the performance on classifying the six skin diseases involved under Erythemato-Squamous diseases varies mainly due to adopted classifiers, the main objective here is to compare the performances of different classifiers on diagnosis of these diseases. The classifiers examined here are support vector machine, discriminant classifier, K-Nearest neighbor and decision tree. Further, we have performed our analysis with two well-known multiclass implementation techniques, i.e., one-vs-one and one-vs-all and compared their performances as well An in-depth comparison has been presented with all classification parameters such as accuracy, sensitivity, specificity etc. This study demonstrates that the most reliable performance has been achieved using support vector machine classifier with one-vs-all approach.