Enhanced Classification method for Large-Scale Medical Data

The medical community has been concerned about how to increase the accuracy of different classification methods with large data that are being generated every day. The traditional KNN method has many limitations, such as dealing with large-scale data, handling uncertain data, and also determining the k parameter for KNN that gives the best result. In this research, three limitations are solved. By using optimized feature selection methods, optimal features are chosen from many irrelevant features. Then, uncertain data that has a conflict with the class label is handled. Finally, the optimal number of k in the KNN method that gives better accuracy is chosen using the ROC curve. The prime objective of this paper is to develop a hybrid optimal model for medical data classification that handles these challenges. The results are evaluated using the accuracy metric. Experimental results show that the enhanced KNN method outperforms the previously used KNN method in used medical datasets