Abstract

Abstract

DETERMINATION OF PERFORMANCE METRIC OF A CLASSIFICATION SYSTEM FOR MEDICAL DISEASES

Abdu MasanawaSagir


Abstract The objective of this study is to determine the performance metric of a classification system on medical diseases. Accuracy, sensitivity and specificity of a classification system were determined. The proposed classifier maximises the correctly classified data and minimises the number of incorrectly classified patterns. For robustness, the proposed method was tested with two different sets of data obtained from University of California at Irvine?s (UCI) machine learning repository. In comparison, the proposed method achieves superior performance when compared to conventional approach ANFIS based gradient descent algorithm as well as to some related existing methods. The software used for the implementation is MATLAB R2016a (version 9.0) and executed in PC Intel Pentium Quad Core N3700 processor with 4.0 GB of RAM. Keywords: artificialneural network, classification, diagnosis.

PDF