| Popis: |
Skin diseases are the most important worrying problems in societies because it affects the patients both physically and psychologically. Skin disease is one of the highly prone to risk with an association of climatic factors around the world. Predicting the skin disease cases associated with influencing factors is the most crucial task. It is very difficult task to identify the appropriate and optimal features for skin disease from the large volume of health sector data available in the world. Previous researchers applied different types of ensemble features selection techniques for the appropriate selection of features which gives highest accuracy with minimum computation time. Classification rate of any algorithm depends on feature extraction techniques and classifier used for classification purpose. Data availability is one of the most significant drawbacks in the health sector if data is available that might be in raw format. Filling missing value and type conversion almost takes 70% of the time. The missing value can be addressed by statistical parameters such as mean, average, and median with stand mechanism in machine learning. The objective of this paper is the selection of significant attributes and removes irrelevant features that affect model performance. The performance of skin disease data can be experimented through K Nearest Neighbor (KNN), Support Vector Machine (SVM) and random forest classifier. The efficiency of the proposed approach is measured through confusion matrix, accuracy, F-measure, precision and Recall. |
