Performance Analysis of Machine Learning Algorithms for Erythemato-Squamous Diseases Classification

Now a days Erythemato-squamous diseases is one of the most common and dangerous skin disease peoples across worldwide are suffering from disease. This is a popular class of dermatology. In this paper various machine learning classifiers has been used for Erythemato-squamous diseases (ESDs) classification and their performance has been compared and analyzed. Features plays an important role in accuracy of classifiers, for this purpose a random forest classifier has been used as feature selection algorithm. These features are compared and best 15 features are selected among available 34 features for classification. Supervised machine learning models are trained and their accuracy, f1-score and time taken for classification has been compared. Logistic regression, support vector machine and K-Nearest neighbor classifier achieves 99% accuracy. Time taken by ensemble learning approaches such as AdaBoost, random forest, light GBM, XGBoost, and extra trees classifiers are relatively higher.